development and evaluation of a 3-dimensional, image-based ...332/datastream... · linear vs....
TRANSCRIPT
The Development and Evaluation of a 3-Dimensional, Image-Based, Patient-
Specific, Dynamic Model of the Hindfoot
A Thesis
Submitted to the Faculty
of
Drexel University
by
Carl William Imhauser
in partial fulfillment of the
requirements for the degree
of
Doctor of Philosophy
August 2004
Copyright 2004 Carl William Imhauser. All Rights Reserved.
ii
Dedications
I would like to dedicate this work to my parents. Thank you.
iii
Acknowledgements
I have so many people to thank for at long last being at the point of printing my
dissertation after one final all-nighter. I would like to express my gratitude to my advisor,
Professor Sorin Siegler, for his steady input and guidance over the past years. I am very
appreciative for his uncanny ability to help reduce a problem to its essential questions and
I hope that some of his considerable abilities have rubbed off on me. I will always be
thankful to him for welcoming me to his home for family dinners, which always included
great discussion, some delicious variety of fish, and some of the best hummus on the
planet. Many other members of my dissertation committee were no less important. I am
extremely appreciative for the advice and guidance of Dr. Brand, including his periodical
subtle reminders asking how much I’ve progressed in writing the dissertation. I am also
very appreciative for Dr. Hillstrom’s and Dr. Okereke’s support, and Dr. Udupa’s resolve
to help define my project so I would not be stuck in school for eternity.
I’d like to thank my mom, dad and entire family for their support while working
towards the doctoral degree. I’ll always remember my dad’s response when I told him I
was thinking about continuing on in school, but was worried about the time it would take
to finish. He said, “I’m 63 years old, 3 years is nothing, do it.” Well, it took 4 years to
finish, but at last it’s over. I will also remember another of his “Imhauserisms” when I
expressed my worries about how difficult it would be. He responded, “No one ever got
anywhere without hard work..” So with these words of wisdom in my head I went on to
pursue my Ph.D. And now on the other side, I can say that both of his statements are true
and I think I have become a better person for heeding his advice. My mom’s support has
iv
been super-human through the years as well and I am incredibly thankful to her. I would
also like to sincerely thank my sisters for their support. Lydia is the only person I know
who the best time to call is 1AM on a weeknight. I am forever thankful for her advice,
not to mention her willingness to allow her monetarily-challenged brother along on
numerous Easter ski trips. I am also super appreciative of my sister Silvia, for almost
always having the door open for me and allowing me the services of her home washing
appliances on a regular basis. Whenever the pile of assorted dirty socks briefs, shorts and
T-shirts was in danger of growing so big it would squeeze me out of my bedroom, I knew
that she would welcome me gladly into her home for some much needed laundry
services. These laundry pilgrimages also provided a much needed stress outlet in the form
of hide-and-go –seek or peek-a-boo with my nieces.
I would also like to acknowledge my lab partner and great friend, Stacie Ringleb.
I feel lucky that our paths crossed at Drexel and I am blessed to be able to call her a
friend. I greatly appreciated having someone to go through most of the Ph.D. journey
together with. Our conversations in the lab and now over the phone were always great
emotional therapy for me. She is the only person I know that is thoughtful enough to send
me an e-birthday card of polka dancing bunny rabbits as well as to continually offer her
no strings attached support and encouragement over the last few months of the final
Ph.D. completion push. I’d also like to express my gratitude to Dr. Guceri and Dr Mun
Choi. I feel like they are primarily responsible for making my doctoral experience at
Drexel great.
Finally I want to express my appreciation to all Drexel graduate students past and
present who I have come to meet over the years. You are my best friends not to mention
v
best sources of social satire: Gaurav Bajpai, Tony Pellgrino, Lauren Shor, Binil Starly,
Jonathon Thomas, Greg Tholey, Dan Marut, Steve Mastro, Chris Kennedy, Dave Lenhert
Karthic Bala, Anna Belllini and all others.
vi
Table of Contents
List of Tables .................................................................................................................... xv
List of Figures .................................................................................................................. xix
Abstract .......................................................................................................................... xxiii
CHAPTER 1. BACKGROUND ......................................................................................... 1
PREVALENCE OF ARTHRITIS................................................................................... 1
Surgical Treatments For Arthritis Of The Hindfoot ....................................................... 2
Ankle and Subtalar Arthrodesis .................................................................................. 2
Total Ankle Arthroplasty ............................................................................................ 3
HINDFOOT INSTABILITY .......................................................................................... 5
Prevalence ................................................................................................................... 5
Detection ..................................................................................................................... 7
HINDFOOT MODELLING ......................................................................................... 10
Description................................................................................................................ 10
Advantages................................................................................................................ 10
Necessity for Rigorous Evaluation ........................................................................... 10
Applications .............................................................................................................. 11
Arthrodesis and Arthroplasty Alignment.............................................................. 11
Arthroplasty Geometry ......................................................................................... 12
Ligament Instability .............................................................................................. 12
Ligament Injury Diagnosis ................................................................................... 13
Ligament Reconstruction ...................................................................................... 15
HINDFOOT MOTION CHARACTERISTICS............................................................ 16
vii
Plantarflexion / Dorsiflexion..................................................................................... 17
Inversion / Eversion .................................................................................................. 17
Internal Rotation / External Rotation........................................................................ 18
Load-Displacement Characteristics .......................................................................... 18
Contributions of the Ligaments to Hindfoot Stability .............................................. 19
The Anterior Talofibular Ligament (ATFL)......................................................... 19
The Calcaneofibular Ligament (CFL) .................................................................. 20
The Posterior Talofibular Ligament (PTFL)......................................................... 20
Interosseous Talocalcaneal Ligament (ITCL) and Cervical Ligament (CL) ........ 20
The Deltoid Ligament ........................................................................................... 21
IMAGE-BASED TECHNIQUES FOR STUDYING HINDFOOT MECHANICS..... 23
NUMERICAL MODELS OF THE FOOT................................................................... 24
Finite Element Models.............................................................................................. 27
3D Rigid Body Dynamic Models ............................................................................. 28
3D Static Equilibrium Models .................................................................................. 28
2-D KinematicModels............................................................................................... 29
Patient-Specific Hindfoot Models............................................................................. 30
CHAPTER 2. PROJECT OBJECTIVES.......................................................................... 31
CHAPTER 3. MATERIALS AND METHODS .............................................................. 34
MODEL DEVELOPMENT TOOLS............................................................................ 34
3DVIEWNIX ............................................................................................................ 34
Bone Surface Identification Software....................................................................... 37
Geomagic Studio 5.0TM............................................................................................. 39
viii
Model Simulation Software ...................................................................................... 42
Equations of Motion - Numerical Development and Solution ............................. 42
Contact .................................................................................................................. 45
Ordinary Bounding Box (OBB) Trees.............................................................. 45
Overlap Test for OBB’s .................................................................................... 49
MODEL DEVELOPMENT PROCEDURE................................................................. 52
Subjects ..................................................................................................................... 52
Development of CAD Representations for Bone Geometries .................................. 52
3DVIEWNIX Segmentation and Post-Processing................................................ 52
Surface Detection Software .................................................................................. 54
Geomagic Studio................................................................................................... 55
Ligament Geometry .................................................................................................. 58
Lateral Ligaments ................................................................................................. 58
Anterior Talofibular Ligament (ATFL) ............................................................ 58
Calcaneofibular Ligament (CFL)...................................................................... 59
Posterior Talofibular Ligament (PTFL)............................................................ 60
Deltoid Ligament .............................................................................................. 61
Posterior Tibiotalar Ligament (PTTL).......................................................... 61
Tibiocalcaneal Ligament (TCL) ................................................................... 61
TibioSpring Ligament (TSL) ........................................................................ 62
Anterior Tibiotalar Ligament (ATTL) .......................................................... 63
Subtalar Ligaments ........................................................................................... 63
Interosseous Talocalcaneal Ligament (ITCL) .............................................. 63
ix
Cervical Ligament (CL)................................................................................ 64
Ligament Mechanics................................................................................................. 65
Collateral Ligament Properties ............................................................................. 66
Subtalar Joint Ligament Properties....................................................................... 67
Cartilage Mechanics.................................................................................................. 67
Derivation of Contact Stiffness............................................................................. 68
Contact Penetration Exponent Derivation ............................................................ 70
FORCING FUNCTIONS AND BOUNDARY CONDITIONS................................... 71
MODEL MEASUREMENTS....................................................................................... 74
Kinematics ................................................................................................................ 74
Finite helical axis rotation and centroidal translation ........................................... 74
Grood and Suntay Parameters............................................................................... 76
Joint Range of Motion and Flexibility ...................................................................... 83
Ligament Strain and Force........................................................................................ 84
Contact Force Magnitude and Location.................................................................... 84
EXPERIMENTAL EVALUATION............................................................................. 85
Subjects ..................................................................................................................... 85
Experimental Tools................................................................................................... 85
Ankle Flexibility Tester (AFT)............................................................................. 85
MR Compatible Ankle Loading Device (ALD) ................................................... 87
MR Scanner .......................................................................................................... 89
Experimental Testing Procedure............................................................................... 89
In vivo Testing....................................................................................................... 89
x
In vitro Testing...................................................................................................... 91
MODEL SIMULATIONS ............................................................................................ 91
Simulation Settings ................................................................................................... 91
Preliminary Simulations............................................................................................ 92
Joint Neutral Position............................................................................................ 92
Evaluation Studies .................................................................................................... 94
In Vivo Model ....................................................................................................... 94
Steady-State Loading ........................................................................................ 94
Cyclic Loading.................................................................................................. 94
In Vitro Model....................................................................................................... 95
Steady-State Loading ........................................................................................ 95
Parametric Studies (Sensitivity Analysis)................................................................. 95
Ligament Orientation ............................................................................................ 95
Ligament Representation ...................................................................................... 96
Linear vs. Non-linear Ligament Mechanics.......................................................... 96
Contact Damping .................................................................................................. 97
Prediction Studies ..................................................................................................... 97
In Vivo ATFL Tear and Combined ATFL/CFL Tear ........................................... 97
In Vitro Model Predictions.................................................................................... 98
CHAPTER 4. RESULTS.................................................................................................. 99
BONE VOLUME COMPARISON .............................................................................. 99
SIMULATION TIMES............................................................................................... 100
PRELIMINARY SIMULATIONS ............................................................................. 100
xi
Hindfoot Neutral Position ....................................................................................... 100
EVALUATION STUDIES ......................................................................................... 102
In vivo Model Evaluation........................................................................................ 102
Static Loading – Kinematics (sMRI Comparison).............................................. 102
Cyclic Loading (Comparison to AFT Load-Displacement Data)....................... 107
In vitro Model Evaluation....................................................................................... 110
Static Loading – Kinematics (sMRI Comparison).............................................. 110
SENSITIVITY ANALYSES ...................................................................................... 127
Ligament Orientation .............................................................................................. 127
Number of Model Ligament Elements ................................................................... 127
Linear Ligament Representation............................................................................. 130
Contact Damping .................................................................................................... 131
PREDICTION STUDIES ........................................................................................... 133
Hindfoot Kinematics............................................................................................... 133
The Effects of Ligament Removal...................................................................... 133
Ligament Elongation, Strain and Forces (In vivo and In vitro Models) ................ 137
The Effects of Ligament Removal...................................................................... 137
Joint Contact Force Magnitudes ............................................................................. 145
Ankle Joint Complex Flexibility Characteristics.................................................... 146
Hindfoot Mechanics in Plantarflexion and Dorsiflexion ........................................ 154
Kinematics .......................................................................................................... 154
Ligament Elongation, Strain and Forces (In vivo and In vitro Models) ............ 156
CHAPTER 5. DISCUSSION.......................................................................................... 160
xii
MODEL DEVELOPMENT........................................................................................ 160
MODEL ASSUMPTIONS AND LIMITATIONS ..................................................... 160
Boundary Conditions .............................................................................................. 160
Ankle Joint Complex Loading Constraints......................................................... 160
Rigidly Constrained Fibula ................................................................................. 162
Anterior Bone and Ligament Constraints ........................................................... 162
Bone (Inter-Cortical) Gaps ..................................................................................... 163
Contact Damping Coefficient ................................................................................. 163
Contact Stiffness ..................................................................................................... 164
Ligament Mechanical Properties ............................................................................ 166
Subtalar Ligaments ............................................................................................. 167
Identification of Ligaments with Broad Attachment Areas.................................... 168
Experimental Equipment ........................................................................................ 168
EVALUATION EXPERIMENTS.............................................................................. 169
In vivo model........................................................................................................... 169
In vitro model.......................................................................................................... 171
Inter-model Variability ........................................................................................... 172
Model Experimental Evaluation Overview ............................................................ 174
SENSITIVITY ANALYSES ...................................................................................... 175
MODEL PREDICTIONS ........................................................................................... 178
Kinematics .............................................................................................................. 178
Inversion / Eversion ............................................................................................ 178
Anterior Drawer .................................................................................................. 179
xiii
Plantarflexion / Dorsiflexion............................................................................... 179
Flexibility................................................................................................................ 181
Ligament Loading................................................................................................... 184
PRELIMINARY CLINICAL SIGNIFICANCE......................................................... 185
Joint Load-Displacement Properties ....................................................................... 185
Ligament Loading................................................................................................... 189
CHAPTER 6. SUMMARY AND CONCLUSIONS...................................................... 190
OBJECTIVE ............................................................................................................... 190
MODEL DEVELOPMENT........................................................................................ 190
SIMULATIONS ......................................................................................................... 192
MODEL EVALUATION ........................................................................................... 193
Comparison of Experimental Measurements and Model Predictions..................... 193
Sensitivity Analyses................................................................................................ 194
Inter-model Variability ........................................................................................... 194
ASSUMPTIONS AND LIMITATIONS .................................................................... 196
MODEL PREDICTIONS ........................................................................................... 198
Kinematics .............................................................................................................. 198
Flexibility................................................................................................................ 199
Ligament Loading Patterns ..................................................................................... 199
PRELIMINARY CLINICAL SIGNIFICANCE......................................................... 200
CHAPTER 7. FUTURE WORK..................................................................................... 201
SHORT-TERM GOALS............................................................................................. 201
LONG-TERM GOALS............................................................................................... 203
xiv
List Of References .......................................................................................................... 204
Appendix A..................................................................................................................... 214
Appendix B ..................................................................................................................... 215
Appendix B ..................................................................................................................... 216
Vita.................................................................................................................................. 217
xv
List of Tables
Table 1: Summary of Existing Foot/Hindfoot Models ..................................................... 25
Table 2: Scene Size and Cell Size for each bone-set ........................................................ 53
Table 3: Ligament non-linear load-strain equation constants[103] .................................. 66
Table 4: Subtalar ligament insertion area ......................................................................... 67
Table 5: Intercortical distance at ankle joint and subtalar joint ........................................ 69
Table 6: The motion constraints imposed on the ankle joint complex for the in vivo and in vitro models in each movement ................................................................................ 72
Table 7: The loads applied for the in vivo and in vitro models and the corresponding rise time (for static loading) or period (for cyclic loading) of loading in each movement.................................................................................................................................... 73
Table 8: Directions for Grood and Suntay joint parameters at ankle joint complex (AJC), ankle joint (AJ) and subtalar joint (STJ) for right (R) and left (L) feet .................... 82
Table 9: Volume comparison of bone geometries obtained from 3Dviewnix and after applying the reduce noise and decimate features in GEOMAGICTM ....................... 99
Table 10: Comparison of ligament lengths and residual ligament forces in MR neutral and closed gap neutral positions for in vivo and in vitro models............................ 101
Table 11: Kinematic changes at hindfoot resulting from closing inter-cortical gaps at ankle joint and subtalar joints for In vivo and In vitro models ............................... 102
Table 12: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to inversion for the in vivo model and the test subject ....................................................................................... 103
Table 13: Grood and Suntay Parameters describing changes in vivo hindfoot model kinematics in between neutral and inverted positions ............................................ 105
Table 14: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to anterior drawer for the in vivo model and the test subject ............................................................................... 106
Table 15: Grood and Suntay Parameters describing changes in bone position from neutral to anterior drawer positions as predicted by in vivo hindfoot model ...................... 107
xvi
Table 16: Comparison of early, late and total flexibility and range of motion [ROM] characteristics of the in vivo model and the patient upon which the model was based.................................................................................................................................. 108
Table 17: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to inversion for the in vitro model and the test subject ....................................................................................... 111
Table 18: Grood and Suntay Parameters describing changes of in vitro hindfoot model kinematics from neutral to inverted positions......................................................... 113
Table 19: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) comparing changes from neutral to anterior drawer for the in vitro model and the test subject ..................................... 114
Table 20: Grood and Suntay Parameters describing changes In vivo hindfoot model kinematics from neutral to anterior drawer............................................................. 116
Table 21: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to inversion with the ATFL sectioned for the in vitro model and the test subject............................................... 117
Table 22: Grood and Suntay Parameters describing changes in in vitro hindfoot model kinematics from neutral to inversion with the ATFL sectioned ............................. 118
Table 23: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to anterior drawer with the ATFL sectioned for the in vitro model and the test subject.................................... 119
Table 24: Grood and Suntay Parameters describing changes in in vitro hindfoot model kinematics from neutral to anterior drawer with the ATFL sectioned.................... 121
Table 25: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to inversion with the ATFL and CFL sectioned for the in vitro model and the test specimen............................ 122
Table 26: Grood and Suntay parameters describing changes in In vitro hindfoot model kinematics from neutral to inversion with the ATFL and CFL sectioned .............. 123
Table 27: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to anterior drawer with the ATFL and CFL sectioned for the in vitro model and the test subject..................... 125
Table 28: Grood and Suntay Parameters describing changes in in vitro hindfoot model kinematics from neutral to anterior drawer with the ATFL and CFL sectioned .... 126
xvii
Table 29: The effect of CFL calcaneal insertion location on ankle joint complex kinematics from neutral to inverted position .......................................................... 127
Table 30: The effect of using multiple force components to represent the TSL and TCL ligament structures on ankle joint complex (AJC) kinematics in eversion ............ 128
Table 31: The effect of using multiple force components to represent the TSL and TCL ligament structures on ankle joint (AJ) complex kinematics in eversion ............... 128
Table 32: The effect of using multiple force components to represent the TSL and TCL ligament structures on subtalar joint (STJ) kinematics in eversion ........................ 129
Table 33: The effect of using multiple force components to represent the deep PTTL ligament on ankle joint kinematics from neutral to inverted position with ATFL sectioned and CFL sectioned .................................................................................. 130
Table 34: The effect of isolated rupture of the ATFL (cATFL) and combined rupture of the ATFL and CFL (cATFL+ cCFL) on hindfoot kinematics in inversion as predicted using the in vivo model ........................................................................... 133
Table 35: The effect of isolated rupture of the ATFL and combined rupture of the ATFL and CFL on hindfoot kinematics in anterior drawer as predicted using the in vitro model....................................................................................................................... 134
Table 36: The effect of isolated rupture of the ATFL and combined rupture of the ATFL and CFL on hindfoot kinematics in inversion as predicted using the in vitro model................................................................................................................................. 135
Table 37: The effect of isolated rupture of the ATFL and combined rupture of the ATFL and CFL on hindfoot kinematics in anterior drawer as predicted using the in vitro model....................................................................................................................... 136
Table 38: The effect of serial ligament sectioning on ligament mechanics in inversion for the intact, ATFL removed (cATFL) and combined ATFL and CFL removed (cATFL+ cCFL) conditions as predicted by the in vivo model .............................. 138
Table 39: The effect of serial ligament sectioning on ligament mechanics in anterior drawer for the intact, ATFL removed (cATFL) and combined ATFL and CFL removed (cATFL+ cCFL) conditions as predicted by the in vivo model ............... 140
Table 40: The effect of serial ligament sectioning on ligament mechanics in inversion as predicted by the in vitro model ............................................................................... 142
Table 41: The effect of serial ligament sectioning on ligament mechanics in anterior drawer as predicted by the in vitro model............................................................... 144
xviii
Table 42: Contact force magnitudes at each hindfoot bone articulation for steady state loading in inversion and anterior drawer in the intact, ATFL sectioned (cATFL), and ATFL+CFL sectioned (cATFL+cCFL) conditions.......................................... 146
Table 43: Flexibility characteristics as predicted by the in vivo model for all motions . 147
Table 44: Flexibility characteristics as predicted by the in vitro model in all motions .. 151
Table 45: Kinematics of the in vivo hindfoot model in plantarflexion / dorsiflexion..... 155
Table 46: Kinematics of the in vitro hindfoot model in plantarflexion / dorsiflexion.... 155
Table 47: Ligament mechanics in plantarflexion and dorsiflexion as predicted by the in vivo model.............................................................................................................. 157
Table 48: Ligament mechanics in plantarflexion and dorsiflexion as predicted by the in vitro model ............................................................................................................. 159
Table 49: Comparison of the volume and principle axes lengths for the hindfoot bones of the in vivo and in vitro test subjects. Cal=Calcaneus, Tal=Talus, Tib=Tibia, Fib=Fibula, Diff=Difference................................................................................... 173
Table 50: Percent difference in ankle joint complex flexibility between the three test conditions for the in vivo and in vitro models and an in vitro experiment (exper)[61]................................................................................................................................. 184
Table 51: Comparison of experimental[123] and predicted in vitro and in vivo percent strain as a function of plantarflexion [-] and dorsiflexiuon [+] angle..................... 185
Table 52: Ankle joint contact locations and force magnitudes [mag] for positions 1-8 of the load-displacement curve shown in Figure 42 ................................................... 188
xix
List of Figures
Figure 1: Segmented MR slice.......................................................................................... 34
Figure 2: Principal Axes of Hindfoot Bones..................................................................... 37
Figure 3: 2D sample of corner identification in marching cubes algorithm..................... 39
Figure 4: Wrapped bone surface representation before and after noise reduction ........... 40
Figure 5: Local bone surface smoothing........................................................................... 41
Figure 6: Triangulated bone surface before (a) and after (b) decimation ......................... 42
Figure 7: OBB Tree Development: Top-down midpoint division of bounded polygon[99]................................................................................................................................... 48
Figure 8: Illustration of separating axis test: L is a separating axis for OBB’s A and B because their half-projections onto L are disjoint..................................................... 49
Figure 9: Comparison of 3DVIEWNIX identified landmark (a) and its location in terms of ADAMS calculated tibia inertial reference frame (b) .......................................... 56
Figure 10: ADAMS-calculated inertial reference frames for hindfoot bones .................. 57
Figure 11: ATFL identification plane, ligament insertion points and placement on hindfoot bones........................................................................................................... 59
Figure 12: CFL identification plane, ligament insertion points and placement on hindfoot bones ......................................................................................................................... 60
Figure 13: PTFL identification plane, ligament insertion points and placement on hindfoot bones........................................................................................................... 60
Figure 14: PTTL identification plane, ligament insertion points and placement on hindfoot bones........................................................................................................... 61
Figure 15: TCL identification plane, ligament insertion points and placement on hindfoot bones ......................................................................................................................... 62
Figure 16: TSL identification plane, ligament insertion points and placement on hindfoot bones ......................................................................................................................... 62
Figure 17: ATTL identification plane, ligament insertion points and placement on hindfoot bones........................................................................................................... 63
xx
Figure 18: ITCL identification plane, ligament insertion points and placement on hindfoot bones ......................................................................................................................... 64
Figure 19: CL identification plane, ligament insertion points and placement on hindfoot bones ......................................................................................................................... 64
Figure 20: Joint gap measurement points ......................................................................... 70
Figure 21: Contact force shown as a function of penetration distance ( ≤ 3 mm) for 107 ≤≤ e (k=0.116).................................................................................................. 71
Figure 22: Definition of anatomical landmarks for Grood and Suntay parameters in ADAMS .................................................................................................................... 77
Figure 23: Grood and Suntay joint axes definitions based on anatomical reference frames for the ankle joint complex, ankle joint and subtalar joint ....................................... 81
Figure 24: Joint flexibility parameters.............................................................................. 84
Figure 25: The Ankle Flexibility Tester ........................................................................... 86
Figure 26: The MRI compatible Ankle Loading Device .................................................. 88
Figure 27: Definition of anatomical landmarks for Grood and Suntay parameters in MRI................................................................................................................................... 89
Figure 28: Anterior view of the in vivo (Left) and in vitro (Right) hindfoot models in the closed-gap neutral position ....................................................................................... 93
Figure 29:Comparison of inverted in vivo model (L) and corresponding MR image data (R) for the intact condition...................................................................................... 104
Figure 30:Comparison of in vivo model (L) and corresponding MR image data (R) for in anterior drawer ........................................................................................................ 106
Figure 31: In vivo experimental (thin lines) and numerical (thick lines) load-displacement curves for inversion [ + ] / eversion [ - ] ................................................................ 109
Figure 32: In vivo experimental (thin lines) and numerical (thick lines) load-displacement curves for internal rotation [ - ] / external rotation [ - ] .......................................... 109
Figure 33: In vivo experimental (thin lines) and numerical (thick lines) load-displacement curves for anterior drawer [ + ] .............................................................................. 110
Figure 34:Comparison of inverted in vitro model (L) and corresponding MR image data (R) for the intact condition...................................................................................... 112
xxi
Figure 35:Comparison of in vitro model (L) and corresponding MR image data (R) in anterior drawer for the intact condition .................................................................. 115
Figure 36:Comparison of inverted in vitro model (L) and corresponding MR image data (R) with the ATFL sectioned .................................................................................. 117
Figure 37: Comparison of the in vitro model (L) and corresponding MR image data (R) for anterior drawer with the ATFL sectioned ......................................................... 120
Figure 38:Comparison of inverted in vitro model (L) and corresponding MR image data (R) with the ATFL and CFL sectioned ................................................................... 123
Figure 39: Comparison of the in vitro model (L) and corresponding MR image data (R) for anterior drawer with the ATFL and CFL sectioned ......................................... 125
Figure 40: Inversion [ + ] / eversion [ - ] load-displacement curve for in vitro model using linear ligament stiffness characteristics .................................................................. 131
Figure 41: The effect of the contact damping parameter (Damping (c) = 0.05, 0.1, 1.0) on the load-displacement characteristics of the ankle joint complex in inversion [ + ] / eversion [ - ] ............................................................................................................ 132
Figure 42: Plantarflexion [ + ] / Dorsiflexion [ - ] load-displacement curves for in vivo model in intact condition ........................................................................................ 148
Figure 43: Inversion [ + ] / eversion [ - ] load-displacement curves for in vivo model in intact, ATFL sectioned and ATFL + CFL sectioned conditions ............................ 149
Figure 44: Internal rotation [ - ] / external rotation [ + ] load-displacement curves for in vivo model in intact, ATFL sectioned and ATFL + CFL sectioned conditions...... 149
Figure 45: Anterior Drawer [ + ] load-displacement curves for in vivo model in the intact, ATFL sectioned and ATFL + CFL sectioned conditions ............................ 150
Figure 46: Plantarflexion [ + ] / Dorsiflexion [ - ] load-displacement curves for in vitro model in intact condition ........................................................................................ 152
Figure 47: Inversion [ + ] / eversion [ - ] load-displacement curves for in vitro model in intact, ATFL sectioned and ATFL + CFL sectioned conditions ............................ 153
Figure 48: Internal rotation [ - ] / external rotation [ + ] load-displacement curves for in vitro model in intact, ATFL sectioned and ATFL + CFL sectioned conditions..... 153
Figure 49: Anterior Drawer [ + ] load-displacement curves for in vitro model in intact, ATFL sectioned and ATFL + CFL sectioned conditions ....................................... 154
xxii
Figure 50: Sample in vitro flexibility test in internal rotation / external rotation........... 170
Figure 51: In vitro model inversion / eversion load-displacement plot overlayed with tibio-talar (Ti-Ta) contact force magnitude. Points marked 1-8 on the load-displacement curve are specifically discussed........................................................ 188
xxiii
Abstract
The Development and Evaluation of a 3-Dimensional, Image-Based, Patient-Specific, Dynamic Model of the Hindfoot
Carl William Imhauser Sorin Siegler, Ph.D.
This study developed subject-specific, three-dimensional dynamic hindfoot models (1 in vivo, 1 in vitro) using 3D stress MRI data. Each model’s ability to capture mechanical phenomena including those of the healthy hindfoot and the hindfoot with ligament injury was evaluated through subject-specific experimental mechanical analyses (using an arthrometer and a stress MRI technique).
Existing software (3DVIEWNIXTM) was incorporated with software developed in-house (marching cubes program) to obtain the subject’s bone surface geometry, collateral and subtalar ligament insertion data. The bone surface data were then imported into a reverse engineering software package (Geomagic StudioTM) to obtain CAD representations for the bone geometries.
The ligaments’ non-linear structural properties were obtained directly from an existing experimental study or were estimated. Contact forces between bones were modeled using cartilage’s Elastic Modulus and an exponential term to imitate its non-linear compression characteristics. The ADAMS 2003TM dynamic simulation software generated and solved the dynamic equations of motion under the forcing functions and boundary conditions.
The in vivo experimental kinematic data were smaller than those predicted by the model. This indicates that surrounding soft tissues excluding the ligaments may decrease joint range of motion. The in vitro model captured the experimental kinematic patterns of the ankle joint complex, but did so by under-estimating ankle joint motion and over-estimating subtalar joint motion. Better knowledge of the ankle joint and subtalar joint ligament structural properties is necessary.
Similar to experimental data, the in vivo and in vitro models’ ankle joint complex had non-linear load-displacement properties in all directions. They are dependent on the contact of the articulating surfaces and ligament constraints. Sensitivity analyses indicated that kinematic changes caused by altering ligament geometry are smaller than changes caused by lateral ligament removal; therefore the model may be sensitive to predicting the changes that occur during ligament rupture.
The models’ assumptions and limitations include differences between the experimental and modeled boundary conditions, exclusion of the cartilage geometry, estimation of the contact damping coefficient, the contact stiffness and penetration exponent, estimation of the subtalar ligaments’ structural properties, generalized non-linear properties for the collateral ligaments, and soft-tissue motion during the experiments. Future work must focus on developing a larger group of patient-specific models so that the output data has sufficient statistical power.
1
CHAPTER 1. BACKGROUND
PREVALENCE OF ARTHRITIS
In the year 2002, nearly 70 million people were living with arthritis according to statistics
provided by the Arthritis Foundation[1]. Osteoarthritis is the most common form of
arthritis, affecting over 20 million Americans, while rheumatoid arthritis, which affects
about 2.1 million Americans, is the most disabling form of the disease[2]. Arthritis is
second only to heart disease as a cause of workplace disability[1]. Various sources have
stated that the foot is involved in 16% to 90% of all cases of rheumatoid arthritis[3, 4].
Arthritis of the ankle joint in patients with rheumatoid arthritis is common but has shown
great variation in occurrence depending on the duration of the disease at time of treatment
as well as the specialization of the clinics to which the patients are referred[4, 5]. After
reviewing a database of 300 patients suffering from this disease, one author found that
the ankle and subtalar joints were affected in 52% of his patients[4]. The patients in this
database had suffered from rheumatoid arthritis for an average duration of 9.5 years[4].
In a database of 1000 patients, 9% of a doctor’s cases involved the ankle joint while the
subtalar joint was affected in 70%[4]. The cases in this study were early in the onset of
the disease[4].
2
SURGICAL TREATMENTS FOR ARTHRITIS OF THE HINDFOOT
Ankle and Subtalar Arthrodesis
The established treatment for severe arthritis in the ankle joint and the subtalar joint is
arthrodesis[5]. There are more than 30 surgical techniques to fuse the ankle joint and the
literature is conflicting and vague as to the most appropriate procedure for patients with
rheumatoid arthritis[5]. Surgical techniques vary because fusing the ankle joint is
demanding due to the small articular contact area relative to other joints such as the knee
and the high contact forces that develop due to the lever arm of the foot[4]. Different
techniques have been used for patients with different diagnoses ranging from
posttraumatic osteoarthritis most commonly to rheumatoid arthritis[4].
There are several common clinical complications resulting from fusion of the ankle or
subtalar joints. These include progressive degeneration of the surrounding joints due to
altered joint mechanics, and bone non-union, where the bony surfaces to be fused do not
heal together and do not become rigid. Although, ankle arthrodesis does relieve pain in
the short term, it is often associated with short-term and long-term functional problems
with stair-climbing, walking on uneven surfaces, and running[6]. Clinicians observed that
ankle arthrodesis would place excessive stress on surrounding joints, leading to
subsequent arthritis[6-9]. For example, radiographs indicated that the tarsal joint
degenerated in 25 of 37 patients in a 1 to 25 year follow-up study of patients receiving an
ankle fusion. It is often necessary to fuse the other degenerating structures once they
become arthritic and painful, which further limits the patient’s mobility[6].
3
The alignment of the fused joint strongly influences the motion at surrounding joints and
the patient’s gait pattern. For example, if the ankle is fused in too much dorsiflexion, the
impact of initial floor contact will be concentrated on a small area of the heel, which can
cause chronic pain[9]. If the ankle is fused in too much plantarflexion, there will be
increased stresses at the joint of the midfoot[9].
Subtalar joint arthrodesis will lead to structural changes in the surrounding joints. For
example, children can develop secondary changes such as forming a ball and socket ankle
joint, secondary to a subtalar fusion[9]. Excessive ligament laxity about the collateral
ligaments of the ankle joint has also been observed[9]. If the joint is fused in a varus
position, the forefoot will supinate and there will be increased stresses on the lateral
collateral ligaments of the ankle joint[9]. Improved results were observed when the
subtalar joint was aligned with a valgus tilt of 5˚. In this position, the ankle joint is
aligned in a stable position, the stress on the lateral collateral ligaments is reduced, and
the weight is more evenly distributed on the plantar aspect of the foot[9].
Total Ankle Arthroplasty
In light of the functional limitations and strong possibility for long-term problems
resulting from ankle arthrodesis, surgeons and companies have explored total ankle joint
replacement (TAR) as an alternative procedure[5-7, 10, 11]. Increased levels of success
with newer implant designs such as the AgilityTM and the STARTM, sparked an increase
in the use of total ankle replacement[5]. These second-generation implants were reported
to provide greatly improved clinical outcome compared to the previous first generation
4
implants (such as the Mayo clinic implant). Nevertheless, rates of failure and clinical
complications are still unacceptably high. For example, in a ten year clinical study of the
AgilityTM conducted on 86 patients, 1 patient required resection of the implant followed
by ankle arthrodesis, 5 patients needed revision surgeries, the tibial component migrated
in 12 patients, and the talar component shifted in 9 patients[11]. 11 of 52 patients
receiving the STARTM implant, which was fixed to the bones with methylacrylate, had
required implant revisions or removal with subsequent ankle arthrodesis after 10 years[6].
Intermediate results using the non-cemented STARTM ankle replacement have been the
most encouraging. After 3.5 years, only 1 of 35 had a revision surgery for malalignment
of the implant components and none showed signs of loosening or subsidence[6].
Failure of second generation TAR’s are related to implant design and to other factors
such as ligament balance, implant alignment and amount of bone resection[6, 12-14]. In
50 ankles implanted with the STARTM, no component migration was observed and all
implants were stable; however, revisions were performed in 7 patients in order to relax
the soft tissue constraints of the medial collateral ligaments (n=2), which were painful to
the patient and restricted motion in dorsiflexion, to lengthen the achilles tendon (n=1), to
relieve impingement of contacting ankle joint bone surfaces (n=1), and to correct tibial
component alignment after stress fracture of the distal tibia (n=1)[6]. After a 5 year
follow-up for the 50 ankles, nearly half (n=23) had pain on the posterior medial side of
their ankle joint, while 32 patients (46%) developed bone hypertrophy in this area[6]. The
hypothesis for the cause of these problems was non-anatomic design of the implant’s
cylindrical talar component[6]. This led to overstress of the posteromedial ankle
5
ligaments, which results in pain and ossification in this area and limited range of motion
in dorsiflexion[6]. Despite the problems associated with implanting the next generation of
total ankle replacements, the surgeons were encouraged by their experience with the
STARTM system[6]. As ankle arthroplasty gains favor, it must be compared with ankle
arthrodesis, the current benchmark treatment for patients with severe osteoarthritis or
rheumatoid disease[5].
HINDFOOT INSTABILITY
Prevalence
Inversion ankle sprains are the most common musculoskeletal injury presented to
emergency rooms[15]. In 2001, 2.3 million people visited physician’s offices because of
an ankle sprain[16]. Furthermore, nearly 25,000 individuals sprain their ankles daily[17].
Severe inversion sprains may lead to ankle instability alone as well as subtalar joint
instability. Up to 20% of all ankle sprains progress to a functional instability of the
hindfoot[18]. These sprains commonly cause injury to the anterior talofibular ligament
(ATFL), and the calcaneofibular (CFL) ligament. Out of 148 patients with symptomatic
chronic ankle instability, arthroscopic examination showed that rupture or elongation of
the ATFL occurred in 86% of patients and of the CFL in 64%. The examinations also
revealed that 40% of patients had injured their deltoid ligament. (Hintermann AMJSM
2002). Chronic inversion sprains also damage the articular cartilage of the ankle joint.
(Hintermann, Tochigi) This occurred in 66% of ankles with lateral ligament injuries and
98% of ankles with deltoid ligament injury[19]. This may lead to the progression of
osteoarthritis or the development of osteophytes[20].
6
An ankle sprain can also damage the structures supporting the subtalar joint, such as the
interosseous talocalcaneal ligament (ITCL) and the cervical ligament (CL)[18, 21, 22].
Subtalar disorders are major causes of chronic ankle pain after an inversion sprain[22].
Between 10% and 20% of all patients that have functional ankle instability also have
subtalar instability[18]. In one study, 32 of 40 (80%) patients had a rupture involving the
subtalar and ankle capsuloligamentous structures[21]. 4 of the 32 (13%) patients had a
surgically confirmed rupture of the ITCL, while 9 of 32 (28%) had a surgically confirmed
tear of the CL[21].
Long-term pain and disability of the ankle joint may occur if an inversion sprain is not
treated properly after the initial injury. Therefore, it is important to detect the location and
extent of ligament injury shortly after the initial incident occurs so that a proper treatment
protocol can be recommended and to minimize the chance for long-term pain and
complications. Many patients with inversion ankle injuries have long-term pain. For
example, 32% of the 648 patients seen in the injury ward of a hospital for ankle inversion
injuries during one year, reported chronic complaints of pain, swelling or repeated sprains
seven years after their injury[23]. 72% of those with continuing pain stated that their
injury impaired daily activities such as participating in sports at the desired level[23].
7
Detection
Although it is vital to properly diagnose the location and extent of ligament damage after
an inversion injury, the detection of mechanical instability in the ankle and subtalar joints
remains difficult. Current clinical approaches such as physical examination, stress
radiography and magnetic resonance imaging (MRI) are inadequate. With a physical
examination, it is impossible to detect the location of the instability (ankle or subtalar
joint). Stress radiographic approaches limit knowledge of hindfoot mechanics to two
dimensions, while static MRI techniques, can detect soft tissue injury, but cannot detect
the instability[15].
The talar tilt and anterior drawer tests are the most commonly used techniques for
diagnosing lateral ligament injury[24-26]. Due to the large variations in range of motion
among subjects and variability among examiners, it is difficult to diagnose specific
ligament involvement, particularly the ATFL. Although the talar tilt test has proven
sensitive to distinguishing injuries to the CFL[24-26] the anterior drawer test may have
low sensitivity in detecting ATFL rupture[24, 27]. Furthermore, these two tests, when
used in combination may still not distinguish between isolated ATFL rupture or
combined tears of the ATFL and CFL[27]. Biomechanical studies support these clinical
difficulties. Sectioning the ATFL caused small but statistically significant increases in the
anterior movement of the talus at the ankle joint[25, 26, 28, 29], which may be difficult to
discern by a clinician[24]. The maximum median increase in talar displacement was 3.1
mm when the foot was tested in a neutral flexion alignment[25]. The ankle joint was most
sensitive to anterior drawer in the neutral laxity region[29] (<2.5N force applied) but it
8
would be difficult for a clinician to consistently apply controlled low level loads to the
foot in order to employ such a test[24]. Total flexibility changes in anterior drawer also
did not correlate well with injury to the ATFL in vitro or to ankle sprains in vivo [29, 30].
The total flexibility increased by 19.3% in 10˚ of dorsiflexion after sectioning the ATFL,
but did not change in the neutral or plantarflexed positions in a study on 12 cadavers[29].
Only 4 of 12 (33%) patients who had experienced 1 unilateral ankle sprain in the
previous 3 months had an increase in total anterior drawer flexibility greater than 18%
compared to their uninjured side[30]. Only 7 of 15 patients who had a history of repeated
ankle sprains (>2 sprains) in the previous decade showed a flexibility increase in anterior
drawer greater than the 18% threshold[30]. Unfortunately, the researchers[30] did not
verify whether the patients had an ATFL injury; therefore, the results of this study cannot
be used to directly calculate the sensitivity or the specificity of anterior drawer flexibility
in detecting ATFL damage.
It is important to develop clinical tests that identify the location of ligament damage with
greater sensitivity and specificity so that doctors can recommend an appropriate course of
treatment (conservative or surgical). The surgeon must be able to diagnose rupture of
specific lateral ligaments and distinguish between collateral ligament damage and
damage to the subtalar joint structures, such as the interosseous ligamant and the cervical
ligament. For example, the internal rotation test may be a more sensitive indicator of
ATFL injury than the drawer test because the ankle joint range of motion increases
greatly (>10˚) after sectioning this ligament[31]. Furthermore, in plantarflexion this
ligament provides 56% of resistance to internal rotation moments[32]. Unfortunately,
9
clinicians use it infrequently[31]. Finding such tests using experimental parametric
studies would be time consuming and costly because the number of specimens needed for
obtaining independent variables to achieve a sufficient statistical power increases as the
number of test variables increase. The most appropriate test for detecting ligament
problems may also be dependant on the orientation of the remaining ligaments, which is
patient specific.
The recently developed “3D Stress MRI Technique” (3D sMRI) solves many of the
problems associated with the approaches mentioned above[33]. This technique assessed
the effects of ligament rupture or sectioning and reconstruction on hindfoot kinematics
and load-displacement characteristics. Even such advanced techniques only evaluate
joint mechanics after an injury or after a surgical procedure. It would be beneficial to the
surgeon to have tools that can be used to determine the effects of various reconstructions
on a patient-specific basis (i.e with regards to the patient’s unique bone and ligament
geometries) This would allow them to plan the most appropriate surgical procedure for
stabilizing the joints of a patient’s hindfoot.
10
HINDFOOT MODELLING
Description
A comprehensive mechanical model of the human hindfoot must be based on three-
dimensional dynamics. Such models use the governing Newton-Euler equations of
motion to solve for the motions in response to applied loads. Furthermore, to provide a
realistic representation of the anatomical structures, these models must incorporate
accurate three-dimensional representations of the bone geometries, and a quantitative
description of the mechanical properties of the surrounding ligaments and their insertion
sites.
Advantages
Researchers can use a rigorously evaluated and confirmed numerical hindfoot model to
perform parametric studies on the effects of ligament injuries and of surgical treatments
described above on joint biomechanics. In general, parametric studies allow the
researcher to isolate one model variable and determine its effects on model output. A
sufficiently evaluated model also enables a large number of studies to be performed in a
much shorter time period and without the expense of studies with patients or cadavers.
Necessity for Rigorous Evaluation
When developing numerical models, it is critical to compare the simulation results with
that of independent experimental data. Independence implies that the model parameters
were not derived, or based in any way, on the experimental data used in the evaluation. In
addition, sensitivity studies must be performed to determine how variations in model
11
parameters affect the final results. Few numerical studies of the foot or ankle joint
complex sufficiently emphasize these aspects of numerical modeling. For example, many
researchers place little emphasis on the role of the ligaments in controlling joint motion.
Anatomical atlases are used to place ligament insertions and guide ligament orientation,
without describing the effect of this methodology on model outcome[34-38].
Furthermore, several studies assigned all of the collateral ligaments the same material
properties[35, 36]. The literature clearly indicates that stiffness of the collateral ligaments
varies greatly, yet the effect of assigning all the ligaments one material property is not
investigated[39]. Many of the models rely on either one simple study, such as axial
impulsive loading[34, 36] or axial compressive loading[38], for model evaluation. Others
provide only qualitative comparisons to the event that they are attempting to simulate[40-
42]. Therefore the range of applications that such models can be used is limited pending
further model evaluation.
Applications
Arthrodesis and Arthroplasty Alignment
The model could be used to predict the effects of position and alignment of joint fusions
and ankle replacement components on the forces in associated ligaments or adjacent
joints and implant contact patterns. With such information, the researcher could
determine the joint fusion position required to minimize abnormal loading of the soft
tissues including the ligaments and the adjacent articulating surfaces. This knowledge
would allow practitioners to minimize the development of arthritic changes at adjacent
joints, abnormal implant wear, as well as ligament laxity and pain[9].
12
Arthroplasty Geometry
Researchers could determine the effects of ankle replacement geometry on joint
kinematics and flexibility characteristics of the replaced and surrounding joints, forces at
the implant articulations and adjacent joints or forces in associated ligaments. Therefore,
the model would aid in designing implant geometries that best recreate normal joint
mechanics and ligament loading patterns before risking failure in patients during clinical
trials. For example, previous design flaws in the STAR’sTM talar component may have
caused abnormal loading of posterior deltoid ligament, which led to posteromedial pain
and ossification at the ankle joint, and limited sagittal plane range of motion[6].
Ligament Instability
Numerical models provide the opportunity to investigate the effects of ligament laxity
and rupture on the motion and flexibility characteristics of the hindfoot joints, ligament
forces, contact locations and contact forces at the articulating surfaces. Damage to the
ligaments alters both the magnitude and distribution of joint forces[43]. The altered
loading pattern may initiate the onset and progression of osteoarthritis[43]. For example,
patients with lateral ligament instability may develop osteophytes and eventually
osteoarthritis[22]. Using a model, the researcher could quantify the loading
characteristics associated with the development of these complications. Based on this
information, doctors could model patients with similar injuries, analyze the changes in
joint loading patterns, and determine whether they are at risk of developing future
complications. Surgical or conservative (bracing) treatments could be applied to the at-
13
risk group, which restore normal joint loading characteristics and reduce their chances of
developing osteoarthritis.
Knowledge of ligament loading patterns enables the researcher to determine structures
that experience increased loading in the presence of partial or total ligament disruption.
This will enable identification of the ligaments that are susceptible to further injury and in
what motions they are most vulnerable (i.e. experience the greatest loading and
displacement). Based on this knowledge, one could develop protective equipment to
most effectively guard the hindfoot ligaments against further damage.
Ligament Injury Diagnosis
Current methods for detecting lateral ligament damage are not sensitive. Manually
forcing the hindfoot into anterior drawer and inversion may not be adequate for detecting
injury to specific lateral ligaments[24, 31, 44]. Clinicians frequently underestimate the
level of ligament injury, often not detecting ATFL rupture, particularly in less severe
cases[45]. For example, 5 of 8 patients diagnosed with grade 2 ligament injury actually
had grade 3 (rupture) of the ATFL, while one did not have any damage to this ligament
(25% sensitivity) as shown by MR evaluation[45]. Stress X-ray techniques also show low
sensitivity in detecting isolated injuries to the ATFL or distinguishing between ATFL and
CFL injuries.[27, 44] For example, a positive anterior drawer test agreed with surgical
confirmation of ATFL rupture in 12 of 20 patients (60% sensitivity)[27]. In addition
positive anterior drawer and talar tilt tests agreed with surgical confirmation of ATFL and
CFL rupture in 38 of 65 subjects (60% sensitivity)[27]. Although MR imaging, due to its
14
excellent soft tissue contrast, shows high sensitivity and accuracy in detecting ligament
injury, it is expensive[46]. In 18 patients, 3D MRI showed 100% sensitivity and 94.4%
accuracy in detecting ATFL rupture and 91.7% sensitivity and 94.4% accuracy in
detecting CFL tear. The gold standard for this study was surgical verification of ligament
integrity[46]. Identifying more sensitive, specific and accurate manual clinical tests could
help the clinician to identify ruptured ligaments without needing costly MR scans.
The basis for such tests is that specific ligament injuries lead to unique changes in joint
motion characteristics. Using a model, researchers could perform parametric studies, in
which they monitor the joint motion characteristics in any type of joint movement. The
movements that result in the largest changes in joint motion characteristics could be
potential tests for diagnosing ligament injury. These studies would be difficult to
complete experimentally due to the large number of test variable and the large number of
specimens that must be tested in order to achieve statistical significance.
Lateral ligament injury may also be associated with natural anatomic variations such as
ligament orientation and bone morphology[47]. Researchers could determine the effect of
ligament alignment on hindfoot range of motion and flexibility characteristics using the
patient specific models. Using this information they could identify groups that are at-risk
of ligament injury and prescribe preventative treatments (use of ankle braces) to reduce
the potential for injury.
15
Ligament Reconstruction
Numerical models could provide a rationale for performing anatomical[48] or tenodesis
procedures[44, 49], using artificial ligament[50] or allograft[51] materials for lateral and
subtalar ligament reconstructions[51]. Anatomical procedures, where the remaining
bands of the ruptured ligament are sewn back together, are the treatment of choice for
lateral ankle instability because it maintains the anatomical alignment of the
ligaments.[48]. These procedures may not be practical because the ligament is no longer
in usable condition or the subtalar joint ligaments (ITCL) are inaccessible. Surgeons must
use tenodesis procedures in these situations. They involve routing pieces of tendon
(typically the peroneus brevis) through the bones of the hindfoot in order to recreate the
supportive function of the lateral ligaments[49]. They may also be performed with
artificial ligament materials[50]. The tenodesis procedures used for treating ankle joint
instability, typically restored ankle joint complex range of motion by reducing range of
motion at the subtalar joint and not fully correcting the instability at the ankle joint[44,
49] Few studies have investigated the biomechanics of isolated subtalar joint ligament
reconstructions[52]. Each tenodesis procedure has been associated with a high incidence
of postoperative pain and some dissatisfaction in the long-term results[44, 49, 52]. These
problems may be associated with the altered joint kinematics and intra-articular loading
conditions, due in part to their non-anatomical recreation of ligament orientation[48, 49].
Given knowledge of the material properties of the replacement tendon or artificial
ligament material, researchers could use a patient specific model to evaluate the effects of
tendon and ligament replacement orientation on joint kinematics, flexibility and intra-
articular forces. Furthermore, when performing the reconstructions, ligament pre-loading
16
may affect the resulting joint mechanics[49]. Researchers could perform parametric
studies to determine appropriate pre-loading levels.
HINDFOOT MOTION CHARACTERISTICS The motion of the ankle joint and subtalar joint is complicated and occurs in all planes. A
model of these structures must have the capability to capture the 3D motion of the joint.
The motion patterns of the hindfoot result from the articulating surface geometry and the
orientation and material properties of the ligaments[32]. Therefore, the model must
include the unique geometry of the patient and accurately reproduce characteristics of the
passive support structures such as ligament strain patterns. Numerous investigators have
described the 3-D motion (kinematics and load-displacement properties) of the joints
experimentally in vivo and in vitro[33, 53-57]. These studies establish the basis for
evaluating and confirming numerical models of the hindfoot; therefore the following
chapter will summarize hindfoot mechanics.
Neither the ankle joint nor the subtalar joint act as an ideal hinge; each contributes to the
total motion across the entire ankle joint complex[53]. Motion occurs in both structures in
all directions (plantarflexion / dorsiflexion, inversion / eversion, internal rotation /
external rotation)[53-56].
17
Plantarflexion / Dorsiflexion
The ankle joint contributes the majority of the range of motion (80%) across the ankle
joint complex in plantarflexion / dorsiflexion with no axial load[53]. The subtalar joint
participates in this motion both when the hindfoot is loaded and unloaded[53, 54, 58]. It
shows small amounts of out-of-plane motion as the hindfoot moves in passive flexion[54,
57]. The calcaneus inverts as the foot plantarflexes[53, 54, 57]. Dorsiflexion coincides
with small amounts of tibia internal rotation (2˚ of internal rotation per 10˚ of
dorsiflexion)[54, 57]. The ATFL, and the anterior tibiotalar ligament (ATTL) undergo
significant elongation (58-87%, 26-51% strain respectively) in planterflexion while the
posterior tibiotalar ligament (PTTL) and the talocalcaneal ligament (TCL) experiences
large elongations (24-46% strain and 11-22% strain respectively) during dorsiflexion[59].
The PTFL also experiences moderate elongation (7 - 17% strain) during both
plantarflexion and dorsiflexion[59].
Inversion / Eversion
The contribution of the subtalar joint to inversion / eversion (73.4% of total ankle joint
complex motion) is greater than that of the ankle joint[53]. Furthermore, the motions of
inversion / eversion are coupled with the motions of internal rotation / external rotation
respectively[53, 55]. Inverting the foot by applying forces typical of a clinical exam
elongates the CFL (24-49% strain)[59]. Furthermore, the PTTL experiences moderate
elongation (9-23% strain) during the same inversion and eversion loading conditions[59].
With no axial load, the lateral ligaments (ATFL, CFL, PTFL) resist 87% of the inversion
18
torque, while the deltoid ligament resists 83% of eversion torque (336-398 N-m for
inversion and eversion)[32]. With an axial load (667 Newtons (N)), the bone articulations
stabilize the hindfoot in inversion and eversion[32].
Internal Rotation / External Rotation
The ankle joint and the subtalar joint contribute equally to the total motion of the ankle
joint complex in internal rotation and external rotation[49, 53]. However, at the extremes
of internal rotation (> 20˚), incremental rotations of the subtalar joint approach twice that
of the ankle joint[53]. Movement of the hindfoot from maximum external rotation to
maximum internal rotation is coupled with substantial hindfoot inversion[53].
The ATFL resists 56% of internal rotation torque when the foot is in 20˚ planterflexion
with no axial load[32]. The deltoid ligament primarily resists internal rotation torque in
foot neutral and 20˚ dorsiflexion.[32] The CFL and PTFL are primarily responsible for
resisting external rotation torque under all axial loading conditions and flexion angles,
while the deltoid ligament plays a secondary supportive role[32].
Load-Displacement Characteristics
The hindfoot has similar load-displacement characteristics in the three anatomical planes
(sagittal, coronal, transverse)[53, 60]. The ankle joint complex is highly flexible under
low forces but flexibility decreases non-linearly at a high rate towards the extremes of
motion[53, 60]. The ankle joint complex is most flexible in the sagittal plane and least
flexible in the coronal plane[53].
19
Contributions of the Ligaments to Hindfoot Stability
The ligaments play an important role in maintaining the passive stability of the
hindfoot[32].
The Anterior Talofibular Ligament (ATFL)
The ATFL supports the ankle joint in internal rotation[31, 32, 49, 53, 59], primarily when
the foot is plantarflexed 20˚ and not axially loaded[32]. Sectioning it caused ankle joint
range of motion to increase by 18% in foot neutral under a torque of 1 N-m[49]. Cutting
it increased hindfoot inversion (5˚)[53] and internal rotation up to 12.1˚[25]. In
plantarflexion, the ATFL elongates substantially (58-87% strain) and assumes a more
parallel alignment with the long axis of the fibula[28], thus making it more susceptible to
injury in this position[61].
The ATFL also supports the ankle joint in anterior translation[29]. It may only play a
partial role in supporting the hindfoot in this motion because it only experienced small
elongation (strain < 8%) when loaded in this position[59]. Sectioning it caused small (<
3mm), but statistically significant increases in ankle joint complex range of motion in this
direction[25]. Neutral zone laxity (defined as the anterior translation of the calcaneus
under a force of 2.5 N) also increased (2mm)[29]. Isolated ATFL sectioning resulted in
increased hindfoot flexibility in this direction[29].
20
The Calcaneofibular Ligament (CFL)
Isolated rupture of the CFL resulted in a significant increase in the range of motion of the
hindfoot in both inversion (15%) (up to 14˚ increase at 3 N-m)[25] and internal rotation
(12%)[53]. Tearing this ligament also caused a significant decrease in the kinematic
coupling of inversion with internal rotation (8%) and vice versa (9%), plantarflexion with
inversion (12%) and internal rotation with plantarflexion (10%)[53]. Sectioning the CFL
caused an increase in the flexibility in all primary directions by approximately 25%. The
CFL resists 50% resistance of inversion torque (336-398 N-m) under no axial load[32].
Its contribution increases to 64% at 15˚ of dorsiflexion[32]. Under inversion loads, the
CFL elongated substantially (24-49% strain)[59]. It also resisted external rotation torques
in all axial loading conditions[32].
The Posterior Talofibular Ligament (PTFL)
The PTFL limits hindfoot range of motion and flexibility during external rotation[53].
These parameters increased 10% and 13% respectively after the PTFL was sectioned[53].
Sectioning the PTFL resulted in up to 36% decrease in external rotation torque when the
hindfoot was held at a constant externally rotated position. It primarily resists external
rotation when the foot is in plantarflexion[32]. It also elongates moderately in anterior
drawer, plantarflexion and dorsiflexion (7-17% strain)[59].
Interosseous Talocalcaneal Ligament (ITCL) and Cervical Ligament (CL)
Subtalar joint instability can occur either in isolation or in association with ankle joint
instability following a lateral hindfoot sprain[62]. Hindfoot instability may persist even
21
following lateral ankle ligament reconstruction[62]. Therefore, the ITCL and the CL may
be important structures in maintaining the stability of the subtalar joint. The ITCL
stabilizes the subtalar joint in supination, but to a lesser extent in pronation[63]. Cutting
either structure resulted in small increases in all clinical rotations (inversion / eversion
(<2.6˚), internal rotation / external rotation (< 2.6˚) and plantarflexion / dorsiflexion
(<1.4˚)) under a constant hindfoot torque of 1.5Nm[62]. The increases, although small in
magnitude, are relatively large in comparison to the total range of movement in the
subtalar joint in all of the clinical motions (> 14%)[62]. The ITCL did not substantially
elongate during any clinical movements (strain < 10%) under loads representative of
those applied during a clinical exam[59]. This indicates that it may be substantially stiffer
than any of the collateral ligaments.
The Deltoid Ligament
The deltoid ligament spans the entire medial aspect of the ankle joint complex, has major
insertions across both the ankle joint and the subtalar joint and primarily resists hindfoot
eversion[32], dorsiflexion[64], and anterior and lateral excursions[65]. Its anatomical
structure varies widely and the origins and insertions of the deltoid’s components are
indistinct; therefore researchers identified its components differently for their mechanical
testing or anatomical characterization studies[59, 61, 65-68]. The primary components
are: posterior tibiotalar (PTTL), anterior tibiotalar (ATTL), tibiospring (TSL)
tibiocalcaneal (TCL) and tibionavicular (TNL). The PTTL always consists of superficial
and deep components, while the ATTL does not always have deep components[66]. The
PTTL contributes to ankle joint stability in dorsiflexion[64] because it undergoes large
22
elongation (24%-46% strain) under loads representative of those applied during a
physical exam[59]. It elongates to a lesser extent (9%-23% strain) in anterior drawer,
eversion and inversion[59]. The ATTL is not present in all people[66] and some consider
it to be indistinct from the TNL and TSL[69]. It supports the ankle joint in plantarflexion
elongating 26% to 51% in this position[59]. It also resists talar anterior translation
elongating 5% to 12% under loads typical of those applied during a physical exam[59].
The TNL is a reinforced fibrous layer of the ankle joint capsule[66]. It supports the ankle
complex during plantarflexion (15-31% strain), anterior drawer (3-7% strain) and
eversion (3-7% strain) during loads typically applied in an exam[59]. The TSL spans the
entire ankle complex attaching along the anterior margin of the tibia’s anterior colliculus
and inserting to the entire medial border of the spring ligament[69]. The TSL and the
TCL are often confused or lumped as the same structures[66]. The TCL lies posterior to
the TSL[69]. It attaches proximally to the posterior aspect of the tibia’s anterior colliculus
and distally to the sustentaculum tali[69]. Sectioning the TCL structure caused an
increase in hindfoot eversion (3.6˚) and very small rotation increases in all other direction
(<1.9˚)[70]. The TSL/TCL structure supports the hindfoot in dorsiflexion (11-22% strain)
anterior drawer and eversion (both 4-11% strain)[59]. Its contribution to hindfoot stability
may be minimal because it has very low stiffness when mechanically tested separately
from the TSL[39].
23
IMAGE-BASED TECHNIQUES FOR STUDYING HINDFOOT MECHANICS
Image-based stress MRI techniques are the means for obtaining patient-specific
geometric data (bones and soft tissues) and mechanical information of joints. These data
are necessary for developing and evaluating numerical models. MR imaging and image
processing tools also eliminate two previous experimental limitations: inaccessibility of
certain joints (ankle joint, subtalar joint) and movement of markers placed directly on the
skin[71]. A custom-built jig moved the foot through the pronation-supination motion and
3DVirewnix derived the relative motion of the individual tarsal bones[71]. 3DViewnix
allowed for bone surfaces visualization and then bone-motion calculation[71].
A new stress MRI technique measured the 6 degree-of-freedom mechanics of the ankle
joint and the subtalar joint under precise loads using an MR-compatible loading
device[33]. This technique allowed the load displacement characteristics of the ankle
joint and the subtalar joint to be measured at discreet loading levels. They obtained the
morphology and kinematic data for the hindfoot under an inversion load and under an
anterior drawer load. The technique resulted in highly reliable measures for the bone
morpohology and a quantitative description of the hindfoot’s architecture[33]. The
results also indicated high left to right symmetry of the healthy ankle and subtalar
joints[33].
24
NUMERICAL MODELS OF THE FOOT
Numerical models to study the foot or ankle complex structures fall into several
categories: 3-D finite element models[34-38, 40-42, 72], 3-D rigid body dynamic models,
3-D static equilibrium models[73, 74] and 2-D kinematic models[75]. The majority of
these studies describe the method (image acquisition, material property considerations,
computational considerations) used to construct the models, but provided minimal
experimental evaluation of model output (Table 1). Researchers mentioned potential
applications of the model such as investigation of joint arthroplasty [34]and ligament
injury[74] but published limited or no results. Few sensitivity analyses investigated the
effect of model simplifications and assumptions such as ligament insertion sites, ligament
material properties and ligament pretension on simulation output parameters (Table 1).
With the exception of one study[40], none evaluated the model output with patient-
specific experimental data. Therefore, these models do not account for anatomical
variations (bone geometry and ligament orientation) or soft tissue material properties
(ligament stiffness) between patients. Furthermore, no studies presented more than one
model; therefore the results cannot have sufficient statistical power for making
conclusions.
25
Table 1: Summary of Existing Foot/Hindfoot Models
Author Model Type Tissues Modelled
Simplifications Experimental Evaluation
Bandak et al.
Finite Element (Non-linear explicit dynamic FE code) LS-DYNA 3DTM
Bones, cartilage, ligaments, retinacula, Achilles tendon, plantar soft tissue
-Linear ligament mechanical properties -Joint kinematics not measured -Non-patient specific evaluation -Rigidly constrained metatarsal bones -Bone geometry correction for non-congruence
-Axial impulsive loading
Beaugonin et al.
Finite Element (Non-linear explicit dynamic FE code) PAM-SAFETM
Bones, cartilage, ligaments, retinacula, Achilles tendon, plantar soft tissue
-Lumped ligament mechanical properties -Estimated ligament insertions and bone geometries -Non-patient specific evaluation -Damping properties of tissues tuned to match simulation results to experiments
-Axial impulsive loading
Camacho, Ledoux et al.
Finite Element Bones, cartilage, ligaments, plantar soft tissue
-Linear ligament mechanical properties -Non-patient specific evaluation -Estimated ligament insertions -Rigidly constrained metatarsal bones
-Tarsal kinematics under axial load
Beillas and Lavaste et al.
Finite Element (Non-linear explicit FE code) RADIOSSTM
Bones, ligaments, cartilage, plantar soft tissue, Achilles tendon
-Lumped linear hindfoot ligament stiffnesses -Estimated ligament insertions
Literature comparison for: -Static and dynamic axial loading, dorsiflexion -Static inversion and eversion
Chen et al. Finite Element MARC K.7.2
Bones, cartilage, ligaments, plantar soft tissue
-Fused medial bones -Fused lateral bones -Lumped ligament material properties -Estimated ligament insertion
Patient-specific plantar pressure pattern
Jacob et al. Finite Element Bones, ligaments, Achilles tendon
-Simplified bone geometries -Fused medial bones -Estimated ligament and tendon insertions
-None
26
Table 1 (continued)
Scott and Winter
3D rigid body dynamic model
Bones, tendons, ligaments, plantar soft tissues
-Fixed 1 degree of freedom ankle, subtalar and metatarsal joints -Rigid transverse tarsal and 2nd metatarsal joints
-Gait data
Kameyama et al.
Rigid Body Spring Modeling
Ligaments, cartilage -Linear ligament properties -Estimated ligament insertion
-None
Bedewi and Digges
3D rigid body dynamic model
Bones -Fixed joint moment-angle properties -Fixed ankle and subtalar joint axes -50th percentile bone geometries
-Axial impulsive loading
Dubbeldam et al.
3D rigid body dynamic model
Bones, ligaments -simplified bone geometries -estimated ligament insertions
-Axial impulsive loading
Salathe Jr., Arangio and Salathe
3D rigid body static equilibrium model (Statically indeterminate structural model)
Bones, tendons, ligaments, plantar aponeurosis
-Rigid subtalar joint -Linear soft tissue material properties -Generic bone inertial properties
-None
Leardini et al.
2D kinematic Ankle joint -Motion limited to passive flexion -Passive motion guided by isometric ligaments -Subtalar motion excluded
-Hindfoot Dorsiflexion
Leardini et al.
2D mechanical model
Ankle joint articulation, hindfoot ligaments
-Sagittal plane load-displacement properties of ankle joint and ligaments -Subtalar motion excluded
-Literature comparison
27
Finite Element Models
Previous finite element models of the foot were based on simplifications that limited their
ability to capture the three-dimensional mechanics; furthermore most models lack
sufficiently broad experimental evaluation, which greatly limits their credibility and
applicability. Some of these simplifications include: bone geometries[34, 41, 42],
ligament mechanical properties and insertion sites[35, 36, 38, 41, 42] and fused hindfoot,
midfoot or forefoot bones[34, 38, 42, 53]. The types of comparison with experimental
evaluation are wide-ranging and include: none[41, 42, 74, 76], existing data from the
literature[72], cadaver experiments[34-36, 38] or patient-specific pressure
distribution[53]. No model results were compared with a database of patient-specific
experimental data. Therefore, no studies documented the ability of their model
development technique to capture unique kinematic and mechanical characteristics due to
variations in subject anatomy.
One use of models is to predict the effects of various surgical procedures or injuries as
described above on joint kinematics and load-displacement properties. Few studies,
though, discussed joint kinematics[35-38, 72], and concentrated primarily on joint
loading levels[34-36]. No studies described joint load-displacement characteristics.
The ligaments of the subtalar joint play important roles in supporting this joint. No study
mentioned the identification and representation of these structures. Several models
assigned all collateral ligaments the same stiffness properties[35, 36, 40], despite their
different mechanical properties[39].
28
3D Rigid Body Dynamic Models
Current 3D rigid body dynamic models of the foot are hindered by limitations similar to
the finite element models, namely: simplified bone geometries, estimated ligament
insertions and limited experimental evaluation. Furthermore, joint geometry and material
properties dictate model dynamics (i.e. joint rotation axes) therefore a numerical model
must capture these features. Several models neglect this requirement by restricting joint
motion to rotation about a predefined axis[77, 78].
3D Static Equilibrium Models
Existing 3D static equilibrium models do not account for moving ankle and subtalar joint
axes or do not include 3D descriptions of articulating surface geometries[73, 74, 79].
The orientation of the subtalar joint changes based on the orientation of the foot[80],
therefore, this could alter moment arm values, which affects the equilibrium equations.
Without experimental evaluation, such an assumption may limit the usefulness of these
models.
The patterns of contact and forces at the articulating surfaces may provide important
information in understanding how they relate to the progression of joint degeneration,
such as in osteoarthritis. 3D equilibrium models based on fixed axes of rotation do not
consider joint geometric characteristics and cannot investigate this aspect of joint
mechanics. They may be used to determine the contribution of support structures
(ligaments, tendons) to joint equilibrium.
29
2-D KinematicModels
Geometric constraints (i.e. – rigid ligaments and joint surfaces) dictated joint motion in
previously developed 2D kinematic ankle joint models[75]; therefore they do not help to
quantify the forces between the articulating surfaces or in the ligaments or the joint’s
load-displacement properties. Researchers developed these models based on observations
that the ankle joint behaves as a single degree of freedom system, with a moving axis of
rotation during passive flexion[75]. The model predicted the planar calcaneus motion,
ligament orientations and lengths, instantaneous axis of rotation and talar surface
profile[75]. Its applications are limited because they cannot account for the 6 degree-of-
freedom out-of-plane, coupled motions occurring at the ankle joint and the subtalar joint.
Recent advances in such 2-D models include response of the ankle joint to anterior
drawer loads and incorporating the collateral ligaments including their material
properties[81]. These simulations remain fundamentally inadequate because they are
planar and do not include the subtalar joint.
30
Patient-Specific Hindfoot Models
Existing imaging technology allows models to be developed using patient-specific
anatomy. The output of such models may be sensitive to unique features in patient
anatomy such as articulating surface curvature and ligament orientation and length.
Therefore the model may capture patient-specific kinematic and mechanical features. No
researchers developed more than one numerical model of the hindfoot on a patient-
specific basis.
Non-invasive stress MR imaging techniques provide the avenue for developing and
evaluating patient-specific numerical models of the hindfoot. Several researchers have
developed methods for analyzing the 3-D kinematics of live joints of the foot based on
image data acquired using magnetic resonance (MR) imaging[33, 71, 82, 83]. 3D sMRI
can measure the 3-D quasi-static load-displacement characteristics of the ankle and
subtalar joints, both in vivo and in vitro called 3D stress MRI (3D sMRI) exists. This MR
technique provides several advantages for understanding joint mechanics over the
previously described experimental studies. It is non-invasive and can be used to assess
3D internal bone kinematics. It can also be used to assess the level of integrity of the
underlying structures. For example, in studies of ligament injuries, 3D sMRI can provide
both visualization of ligament damage and the effect of this damage on the mechanics of
the joints 3D sMRI uses a MR scanner compatible 3D positioning and loading linkage to
hold the hindfoot loaded with precise loads while it is being scanned. Then, using an
image processing technique[71] the bone morphology, architecture and joint kinematics
is determined[33].
31
CHAPTER 2. PROJECT OBJECTIVES
The main objective of this study is to develop a subject specific, three-dimensional
dynamic model of the hindfoot using 3D sMRI data and evaluate its ability to capture a
wide range of mechanical phenomena including the mechanics of the non-pathologic
hindfoot and the mechanics of the hindfoot with ligament injury.
Aim #1: Identify appropriate anatomic information from MRI.
1A. Identify on MRI patient-specific origins and insertions for the collateral ligaments of
the hindfoot and the interosseous and cervical ligaments.
1B. Identify on MRI the insertion area for the interosseous and cervical ligaments.
Aim #2: Combine existing image processing and CAD modeling software to transform 3-
D hindfoot anatomical information, obtained from MR images of the hindfoot, to CAD
representations on a patient-specific basis.
32
Aim #3: Identify appropriate mathematical formulations for the structural properties of
the hindfoot ligaments and cartilage.
3A. Use data documented in the literature to represent the structural properties for the
collateral ligaments.
3B. Scale the average elastic modulus of the collateral ligaments by the insertion areas of
the interosseous and cervical ligaments to represent their structural stiffness properties.
3C. Obtain the local contact stiffness of the articulating surfaces by using the
compressive modulus of cartilage as documented in the literature.
Aim #4: Develop a dynamic model incorporating anatomic information, the CAD
representations for the bones and the structural properties of the ligaments and
articulating surfaces in the ADAMSTM modeling software.
33
Aim #5: Evaluate the model output by comparing it to three types of independent
experimental data.
5A. Compare the experimental ankle joint complex flexibility and range of motion data of
one normal subject obtained using a six-degree-of-freedom mechanical linkage, the
Ankle Flexibility Tester (AFT), to the same quantities calculated with the dynamic
model.
5B. Compare the kinematic data of the hindfoot joints obtained from a stress MR study of
one normal subject, to the same quantities calculated with the dynamic model.
5C. Compare the experimental kinematic data of one cadaver in the intact condition and
with two simulated injuries (anterior talofibular ligament sectioned and anterior
talofibular ligament + calcaneofibular ligament sectioned) to the same quantities
calculated with the dynamic model.
34
CHAPTER 3. MATERIALS AND METHODS
MODEL DEVELOPMENT TOOLS
3DVIEWNIX
3DVIEWNIX is an image processing and visualization software system. It was used to
process all of the collected MR image data and identify ligament insertions. The MR
image processing consisted of several steps described below:
Step 1: Segmentation
The 3D images acquired for neutral, inversion, and anterior drawer configurations
for a given subject are denoted by IN, II, and IA respectively. The boundary of each bone
in each 2D MR slice of each of IN, II, and IA is identified (Figure 1).
Figure 1: Segmented MR slice
The interactive process used for this purpose is referred to as “live-wire”[84]. The user
initially picks a point p0 on the bone’s boundary in the 2D slice using the mouse-pointing
35
device. A “live-wire” path is created between this point and any subsequent position p1
of the pointer by the computer, which represents the best among all possible paths
ints. If p1 is moved close to the boundary and if it is not too far off
the ankle complex such as the talus and calcaneus.
long bones (tibia and fibula) are truncated in a uniform way
erence frame that is used to trim the bones so that they produce the
images I’NB, I’Ib, and
I’Ab co
between the two po
from p0, the live wire snaps onto the boundary. Subsequently, p1 is deposited via a mouse
click, which now becomes p0 and the process continues. For most bones, three to seven
points selected along the boundary are adequate to delineate the entire boundary. The
live-wire method uses dynamic programming and the “live wire” is displayed in real-
time. The output of this step is binary images INb, IIb, and IAb corresponding to IN, II, and
IA.
Step2: Iso-shaping
Two of the bones comprising the hindfoot, the tibia and fibula, are long bones, which
appear only partially within the field of view used for MRI scanning. This presents a
problem with subsequent analysis of the 3D binary images for these bones since our
methods require that the entire bone is covered within the field of view. This is obviously
not a problem for the small bones in
To address this problem, the
with a method called iso-shaping[85]. Iso-shaping identifies a “kernel” called shape
centers representing a set of key points in the partial anatomic structure. These points are
present in all images acquired for the tibia and fibula in different positions. These shape
centers provide the ref
same shape in each position. The output of this step is new binary
rresponding to INb, IIb, and IAb.
Step 3: Surface construction
36
Sur
tained by
principal component analysis of the surface points [71, 86] expressed in the scanner
coordinate system, the length of the intersection of each principal axis with the surface of
the bone, and the eigenvalues associated with each principal axis. The eigenvalues are
comparable to the principal geometric moment of inertia (assuming a uniform mass
distribution inside the surface). A display of the principal axes for each of the bones of
the hindfoot along with the surface of the bone is shown in Figure 2. In these displays,
the principal axes are drawn from the geometric centroid of the bone’s surface.
faces are created and displayed after the binary images produced in the previous
step have been interpolated and filtered with a smoothing Gaussian filter [86]. The
purpose of filtering is the estimation of surface normals that are, as much as possible, free
from digital artifacts.
Step 4: Estimation of Morphological and Architectural Parameters
These parameters are estimated from the surfaces output in Step 3 for each bone
in each configuration by using well-established methods.[71, 86-88]
The morphological parameters computed for each bone include the location of the
geometric centroid of the bone’s surface in the scanner coordinate system (this frame is
attached to the originally acquired volume image), the volume enclosed by the surface of
the bone, the direction of each of the three principal axes of the bone (ob
37
Figure 2: Principal Axes of Hindfoot Bones
Bone Surface Identification Software
obtained from the segmented 2D MR slice data (BIM files) using software
developed in-house by Dr. George Grevera (Medical Image Processing Group,
identifying the surface coordinates from the BIM files. It uses gray data as input
(thresholding) prior to surface triangulation. To employ the expertly segmented
BIM files (and to bypass the inherent thresholding) for each hindfoot bone, they
in the PGM format so that the above program could be used to triangulate the
Cartesian coordinates describing the outer surface of each hindfoot bone were
University of Pennsylvania). The software included several algorithms for
(in the form of PGM files) and performs a rudimentary segmentation
had to first be converted to gray IM0 data. This was then converted to gray data
bone surfaces using an implementation of the Marching Cubes method[89].
38
The boundaries of the 2D slices in PGM format are identified using the marching
cubes algorithm. This technique is used in surface rendering to construct a
triangulated surface from a 3D field of values. It locates the surface corresponding
to a user specified value and creates triangles representing the surface[89]. The
surfaces are located using a divide and conquer approach. The algorithm creates a
logical cube from eight pixels; adjacent slices contribute 4 pixels each to form the
8 vertices of a cube[89]. Marching cubes then determines how the surface
intersects the cube and then repeats the procedure on the next cube. The marching
cubes algorithm tests the corner of each cube (or voxel) in the scalar field as being
either above or below a given threshold. This yields a collection of boxes with
classified corners.[90]. Since the cube is composed of 8 vertices and each vertex
can be insid the
ube[89]. Two symmetry properties reduce the number of cases to 14. Each cube
is given its own 8-bit index, 1 bit for each cube vertex[89]. The bits in each index
are then assigned a value based on which vertices are inside or outside of the
surface. As an example, if the surface includes one vertex of the cube, 1 triangle is
defined by three edge intersections[89]. Other patterns can produce multiple
triangles describing how the surface intersects the cube[89]. The triangle vertex
coordinates describe the surface[89]. The 2D analog would be to take an image,
and for each pixel, set it to black if the value is below some threshold, and set it to
white if it's above the threshold. (Figure 3[90] below)
e or outside the surface, there are 28 ways the surface can intersect
c
39
Figure 3: 2D sample of corner identification in marching cubes algorithm
Geomagic Studio 5.0TM
The Geomagic Studio software can convert and manipulate the surface coordinate data
obtained as output from the marching cubes algorithm to standardized solid CAD model
formats (i.e. IGS, STEP and STL). The bone geometries were processed in Geomagic
Studio in the following steps: 1) global noise reduction, 2) point wrapping, 3) local
surface smoothing and 4) point decimation.
In the global noise reduction step, statistical methods reduce geometric abnormalities
about the entire bone surface (Figure 4). The data output from the Marching Cubes
algorithm was arranged so that the bones had a staircase structure as shown on the left in
Figure 4. The bone model on the right in Figure 4 shows the bone surface after noise
reduction. The point wrapping process converts the point cloud data to polygons
describing the bone surface.
40
Before After
Figure 4: Wrapped bone surface representation before and after noise reduction
ollowing the wrap phase e in the Polygon F , the model is converted to a closed volum
phase. This phase has several surface editing features that allow any non-anatomical
geometric artifacts (as determined by the user) to be removed such as spikes on the bone
surface. The defeature command removes small spikes on the bone surfaces. This
command refits selected regions with a new triangulated surface. For areas with greater
surface irregularity as shown in Figure 5a at the medial talar tuberosity, it is not possible
to use this function and the delete polygons tool was used to remove the selected section.
First the appropriate area is highlighted as shown in Figure 5b. The highlighted area is
then deleted, which leaves holes in the surface representation of the bone as highlighted
in green in Figure 5c. The fill holes tool was used to refit this area with a new polygon
surface as shown in Figure 5d highlighted in red.
41
Figure 5: Local bone surface smoothing
This operation constructs a polygonal structure to fill the hole, and both the hole and the
surrounding region are remeshed so the polygonal layout is organized and continuous.
The spikes tended to occur at locations where the bone geometry was not well defined
such as ligament insertion points. They did not occur in regions with smooth geometries
such as the articulating surfaces. Finally, the decimate polygons tool is used to reduce the
number of triangles represe
a b
c d
nting the bone surface. This tool was run in shape
reservation mode, which ensured that the objects overall shape was preserved. p
ADAMS’TM simulation times are directly related to the number of points describing bone
surfaces, therefore it is important to use the lowest amount possible while maintaining
bone geometry. Each bone was described with 3,000 triangles (6,000 points). Figure 6
below shows the triangulated surface of the calcaneus represented with 265,366 triangles
(a) and the surface decimated to 3,000 triangles (b). The initial volume of each bone was
42
compared before and after the steps described above to monitor their overall effect on the
bones models.
a b
Figure 6: Triangulated bone surface before (a) and after (b) decimation
Model Simulation Software
Equations of Motion - Numerical Development and Solution
The ADAMS software is a 3D rigid body dynamic analysis software package. It uses a
edictor-corrector numerical algorithm to solve the dynamic equations based on the pr
motion time history and current motion trajectory. This formulation is suitable in
circumstances that involve rapid increases in forces due to contact, or rapid changes in
bone position in response to low applied forces due to the geometric non-linearity of the
articulating bone surfaces. The dynamic analysis involves developing[91] and then
integrating[92-94] the non-linear ordinary differential equations of motion as summarized
in the following paragraphs.
Ordinary differential equations (ODE’s) are characterized as stiff or non-stiff. ODE’s are
stiff when they have eigenvalues with large differences meaning the system has both high
and low frequencies. Typically, the high frequencies are overdamped; therefore the
43
system does not vibrate at these frequencies[93]. If these frequencies are not overdamped
then they are active and the system becomes non-stiff. Non-linear ODE’s may be stiff or
on-stiff at different points in time. n
ADAMS formulates the system equations as shown below[93, 95]. :
( ) 0, =−+ qqFAqM TTq &&& λφ Equation 1
( ) 0, =tqφ Equation 2
M is the mass matrix of the system. q is the set of coordinates representing
displacements.φ is the set of configuration and applied motion constraints. is the set
of applied forces and gyroscopic terms of t s. is the matrix that projects
F
he TAinertia force
the applied forces in the direction q . qφ is the gradient of the constraints at any given
state. Equation 1 is a 2nd order ODE and equation 2 is an algebraic equation. ADAMS
uses previously developed DAE integrators, including the GSTIFF I3 and SI2
converts equations 1 and 2 to 1st order Differential Algebraic Equations (DAE) and then
formulations, to solve the system of equations[93, 94].
Integrators are classified as stiff or non-stiff based on their ability to handle these types of
ODE’s and DAE’s. The time step in stiff integrators is limited by the highest active
frequency. For non-stiff integrators, it is limited by the highest frequency in the system,
therefore non-stiff integrators will not solve stiff systems effectively[93]. Many
mechanical systems, including biomechanical ones, are stiff, therefore ADAMS uses a
specially developed algorithm to solve such problems, GSTIFF[94]. Integration involves
two phases: a Prediction and then a Correction[92-94]. When taking an integration step,
44
the integrator fits a high order polynomial, through the past values of each system state
and then extrapolates to find the value at the next time step. The GSTIFF integrator uses
a Taylor’s series to develop this polynomial[92]. This is an explicit process and considers
only past values[92]. Therefore, the predicted value may not satisfy the equations of
motion, especially for systems undergoing rapid changes.
The corrector formulae are an implicit set of difference relationships relating the
of the states at the current time to the values of the states[93]. The non-linear derivative
ODE’s are transformed to a set of non-linear, algebraic difference equations in the system
states[93]. For example, the Backward Euler technique represents a differential equation
as first order difference relationship. Given a set of ODE’s of the form shown in equation
3[93]:
( )tyfdtdy ,= Equation 3
Backward Euler uses the difference relationship:
+11+ += nn yh& Equation 4 n yy
Where ny is the solution at time, ntt = ; h is the step size, and 1+ny is the solution at time,
1+nt .
This is an implicit method for solving ODE’s because 1+n is on both sides of the
equation. ADAMS uses a quasi-Newton-Raphson Algorithm[92, 93] to solve the
difference equations and obtain values for the state variables. This algorithm requires a
matrix of the partial derivatives of each system equation with respect to the system
variables, the Jacobian matrix[93]. This matrix is obtained by linearizing the system
equations about an operating point and it is a function of the integration order and the
45
step size[93]. After the corrector has calculated the Jacobian, the integrator estimates the
error in the solution. If the error is greater than a given threshold, the solution is rejected,
and the system equations are solved with a smaller time step. If the error is less than the
prescribed tolerance, the integrator accepts the solution and takes a new time step[93].
When system dynamics change rapidly, (i.e. large forces turning on or off due to contact
or friction) the premise for using the predictor is violated and it will give faulty values,
hich may cause the corrector to fail[93]. Therefore, it is vital to choose a small enough
integration step size to capture changes in system dynamics. It is also vital to avoid using
programming features such as a binary on-off logic that may lead to such changes.
Contact
In order to determine contact between rigid bodies, ADAMS uses the RAPID
Interference Detection Algorithm[96]. This algorithm computes efficient and exact
interference detection between complex polygons undergoing rigid body motion[96].
This algorithm accomplishes two main tasks: 1. divides the polygons describing the
geometric surfaces into sets of oriented bounding boxes, called OBB Trees[96, 97] and 2.
[96, 97].
Ordinary Bounding Box (OBB) Trees
w
TM
tests pairs of OBB’s for overlap using the separating axis theorem
The OBB Trees are built by computing the convex hull of the vertices of the triangles
comprising one body and then recursively partitioning the convex hull into the smallest
possible group of triangle vertices[97]. The convex hull is the smallest convex set
containing all the points[96]. This convex hull is divided using a top-down approach,
46
where all polygons are recursively grouped and sub-divided until the smallest possible set
of nodes is left. The convex hull is divided along the 3 eigenvectors of the covariance
atrix[96]. These eigenvectors are mutually orthogonal and 2 are the axes of maximum
]. Using this formulation the covariance matrix, , is
simplified to the following form,
m
and minimum variance; therefore they are aligned with the longest and shortest axes of
the hull[96]. The convex hull is sampled infinitely densely, which normalizes for the size
and distribution of the triangle vertices describing the object’s surface[96]. This allows
the orientation of the bounding box to be independent of dense concentrations of vertices
along an arbitrary axis[96 C
( )( )[ ]ik
ij
i
k
i
j
i
k
i
j
ik
i
k
i
k
ij
i
j
i
j
n
i
ijk rrqqpprqprqpm
nC +++++++= ∑
=1241 , 3,1 ≤≤ kj
Equation 5
where, is the number of triangles and n ,µ−= iipp ,µ−= ii
qq .µ−= iirr Each
represents a vector and are the elements of the covariance matrix. Furthermore,
, the area of triangle is,
13× jkC
im thi
47
im = ( ) ( )iiii prpq −×−21 , Equation 6
and the mean point, µ , of the convex hull is
( )iiin
ii rqp
mn++= ∑
=1
161µ . Equation 7
The RAPID algorithm splits the OBB’s along their longest axis[96]. The subdivision
coordinate along this axis is the mean point, µ , of the vertices[96]. When the longest
axis can no longer be split, the next longest is split, then the shortest one. When each axis
is split as many times as possible, using the average criterion, the group is considered
indivisible and the algorithm is complete[96].
The bounding box encloses the extremal points of the 2D polygon and then the box is
divide ]. At
each division the subsequent bounding boxes can have different orientations.
Furthermore, since the objects are rigid, the OBB tree must only be developed once
during pre-processing[97].
d at its midpoint along the longest dimension as shown in Figure 7 below[98
48
Figure 7: OBB Tree Development: Top-down midpoint division of bounded polygon[98]
49
Overlap Test for OBB’s
After the OBB’s have been partitioned into the smallest possible set of vertices, an
algorithm om two different objects for overlap. The overlap test is
based on the separating axis theorem, which states that, “two convex polytopes are
if and only if there exists a separating axis orthogonal to a face of either polytope
or orthogonal to an edge from each polytope[96].” Since each box has 3 unique face
tests the OBB pairs fr
disjoint
orientations and 3 unique edge directions there are 15 possible separating axes that must
be tested. For example, assume we have an ar e wish to determine if
Figure 8: Illustration of separating axis test: L i r OBB’s A and B because their half-projections onto L are disjoint.
L
bitrary axis, L, and w
it is a separating axis (Figure 8). Each bounding box, A and B, forms an interval on the
axis, L, when projected onto it. If the intervals do not overlap then L is a separating axis,
but if they do, then L may be a separating axis and the boxes, A and B, may be disjoint.
This test only excludes potential separating axes.
s a separating axis fo
T●L
rA
A
a1A1a2A2
T
rB
B
1
2
b B1
b B2
50
This approach can be expanded to 3 dimensions as summarized from the literature in the
formulation below[96]. First, assume that there are two OBB’s, A and B, where the
matrix, [ ]R , and the vector, , describe the orientation and position of OBB B with
respect to OBB A. The axes of each OBB are defined with the unit vectors,
T
iA and Bi, for
and the half-dimensions along each box axis are and for . The
length of the projection of box A’s half-radius, r, onto an axis, L, is:
3,2,1=i ia ib 3,2,1=i
∑=
•=3
1iiA ar LAi Equation
where the unit vector,
8
iA , is scaled by the length of each box’s half radius, , and
he distance between the midpoint of the
OBB’s with respect to the possible separating axis, L, is
i
dotted with the possible separating axis, L. This approach can also be used to determine
the radius of the interval of OBB B. If we assume that the potential separating axis’
origin is through the center of OBB A, then t
a
LT • and the intervals are
disjoint iff:
∑∑==
•+•>•3
1
3
1 ii
ii ba LBLALT ii Equation 9
Equation 9 can be simplified if L is a box axis or cross product of box-axes. For example,
if we define, 21 ABL ×= , then the second term in the second summation shown in
Equation 9 becomes:
( )212 ABB ו2b Equation 10
and then after applying an identity and simplifying, it becomes:
32 BA •2b . Equation 11
51
[ ]R , Remember that the dot product of A2 and B3 is R23 of the direction cosine matrix,
describing the orientation of OBB B with respect to OBB A, Equation 11 can be further
simplified to:
23R2b . Equation 12
ssion remains:
After we simplify all the terms in Equation 9, (some terms disappear because of a cross
product between two of the same vectors) the following expre
>• LT 13R1a + 11R3a + 23R2b + 22R3b . Equation 13
All of the 15 axes that must be tested for overlap can be simplified by representing the
separating axis, L,in this manner. With this approach, an OBB that contains only one
polygon and, therefore has no thickness further simplifies the overlap test expressions
above and poses no computational difficulties.
52
MODEL DEVELOP
s
MENT PROCEDURE
Subject
Image data were collected from 1 8.
examined th jec esting t no foot ies
existed. Data were also collected from cadaver leg the
k ed f an room ore testing. ded
the basis for d opi ating el.
Development of CAD Representations for Bone Geometries
hea ge 4lthy volunteer, a An eon orthopaedic surg
e sub t prior to t and verified tha or ankle patholog
an unembalmed disarticulated at
nee stor rozen d thawed to temperature bef These data provi
evel ng and evalu the hindfoot mod
se ns e d
above in the followin se ss
running the surface detection software, and using Geomagic Studio for global and local
oothing and point decimation.
CAD repre ntatio for the bon geom elopeetries were dev using bed the tools descri
g order: gmentation and post-proce ing using 3DVIEWNIX,
bone surface sm
3DVIEWNIX Segmentation and Post-Processing
The MR scans of the neutral positioned hindfoot were processed as described in the
section titled “3DVIEWNIX” above. The image data had already been segmented from a
previous study so this step was not repeated. After segmentation the image data had the
size (scene size) and resolution characteristics (cell size) described in Table 2 (rows
labeled ‘Pre’).
53
Table 2: Scene Size and Cell Size for each bone-set
Bone Interp. Size In vivo In vitro
Interp=Interpolation, pre=before interpolation, post=after interpolation
Tibia
Pre
Post
Scene size Cell size
Scene size
512 x 512 x 258
Cell size
.35mm x .35mm x .35mm 512 x 512 x 258
512 x 512 x 240
.35mm x .35mm x .35mm
35mm x .35mm x .35mm 512 x 512 x 240
.35mm x .35mm x .35mm
Fibula
Pre
Post
Scene size Cell size
Scene size
512 x 512 x 258
Cell size
.35mm x .35mm x .35mm 512 x 512 x 258
512 x 512 x 240
.35mm x .35mm x .35mm
.35mm x .35mm x .35mm 512 x 512 x 240
.35mm x .35mm x .35mm
Talus
Pre
Post
Scene size Cell size
Scene size
512 x 512 x 41
Cell size
.35mm x .35mm x 2.1mm 512 x 512 x 258
512 x 512 x 41
.35mm x .35mm x .35mm
.35mm x .35mm x 2.1mm 512 x 512 x 240
.35mm x .35mm x .35mm
Calcaneus
Pre
Post
Scene size Cell size
Scene size e
512 x 512 x 41
Cell siz
.35mm x .35mm x 2.1mm 512 x 512 x 264
.35mm x .35mm x .35mm
512 x 512 x 41 .35mm x .35mm x 2.1mm
512 x 512 x 240 .35mm x .35mm x .35mm
m3. Interpolation increased the number of
slices in a scene and the cell size as summarized in the pre and post interpolation rows for
each bone (Table 2, ‘Post’). The tibia and fibula were previously interpolated to perform
the iso-shaping operation; therefore they were not processed again and maintained the
same scene size and cell size.
The talus and calcaneus have unequal scanning resolutions (cell size) in each dimension
(Table 2). The slice separation distance (2.1 mm) is much greater than the pixel size
within an individual slice (.35 mm). A linear shape-based interpolation algorithm[99]
increased discretization along the slice direction. This resulted in image data with
isotropic discretization: 0.35 x 0.35 x 0.35 m
54
Following interpolation, a 3D Gaussian filter with a standard deviation of 2 was used to
1 and 256. The
nal step in 3DVIEWNIX is thresholding each of the interpolated bone image slices to
bone surface vertex coordinates were in terms
f the MR scanner reference frame and were scaled by the product of the largest cell
dimension and the largest scene dimension. In order to maintain the relative position and
size of each bone, the bone surface coordinate data was scaled back to its original values
with a simple MATLAB program.
smooth the bone surfaces. This filter is a convolution operator that acts to blur the image.
The degree of smoothing is determined by the standard deviation of the Gaussian.[100]
The output voxel gets a value that is the weighted sum of its surrounding input
voxels[86]. These voxels are gray scale and can have a value between
fi
obtain a binary representation for each MR slice. This assigns each slice’s gray scale
voxels a binary value (black or white) based on the chosen threshold value. The threshold
value was set to 116.
Surface Detection Software
Following interpolation, filtering and thresholding in 3DVIEWNIX, each bone’s BIM file
was processed using the surface detection software (Marching Cubes algorithm) and
output to a TXT file. The triangulated 3D
o
55
Geomagic Studio
Next, the TXT file containing the bone surface coordinate data was imported to the
Geomagic software. Each bone was processed using the steps described above and saved
as closed solids in STL format. The bone volumes calculated in 3DVIEWNIX were
compared to the volumes calculated after CAD modeling in Geomagic.
The individual bone models were then imported into ADAMS using its Import STL
feature. A qualitative check was made to ensure that the CAD representations had similar
inertial properties to their 3DVIEWNIX counterparts as follows: prominent landmarks
were identified using 3DVIEWNIX’s Measure feature (Figure 9a); then the landmark
coordinates were transfo
ference frame using the MATLAB program in Appendix A; then these inertial
rmed from the MR reference frame to the bone’s inertial
re
coordinates were used to define the position of a marker within the CAD model’s inertial
reference frame in ADAMS (Figure 9b) .
56
Figure 9: Comparison o
If the marker location
the CAD representati
segmented and recon
showed consistently c
the corresponding pos
vivo subject and its in
the simulation (talus, c
a b
x
f 3DVIEWNIX identified landmark (a) and its location in terms of ADAMS calculated tibia inertial reference frame (b)
did not qualitatively change between 3DVIEWNIX and ADAMS,
ons for the bones were considered to accurately represent their
structed versions. The CAD models of the tibia, talus and fibula
lose agreement between the landmark picked in 3DVIEWNIX and
ition in ADAMS. The CAD representations for the bones of the in
ertial axes are shown below (Figure 10). Each bone that moved in
alcaneus) was assigned a mass of less than 0.25 kg[72].
57
Figure 10: ADAMS-calculated inertial reference frames for hindfoot bones
58
Ligament Geometry
Ligament insertions were identified by applying 3DVIEWNIX’s manipulate and measure
features to each subject’s reconstructed and assembled hindfoot bones. The collateral and
subtalar ligaments were identified by first intersecting the assembled hindfoot structure
with a plane that was oriented in a manner best suited for identifying the course of the
ligament. Next, the measure tool was used to obtain the location of the ligament insertion
points in terms of the MR scanner coordinate frame. Then the insertion coordinates were
tr
rogram (APPENDIX A). Typically, the ligaments have lower signal intensity than the
ntified by their darker appearance on the image slices.
ansformed from the scanner frame to the inertial frame of a bone using a MATLAB
p
surrounding tissues and can be ide
The method for identifying each ligament is described below.
Lateral Ligaments
Anterior Talofibular Ligament (ATFL)
Aligning a transverse plane that transected the ankle joint at the level of the head of the
lus as shown in Figure 11 identified the ATFL. The plane was then moved up or down
along the tibial long axis until identifying a darker structure running from the anterior
distal portion of the fibula to the medial head of the talus. Figure 11 are images of the
slice plane and the corresponding ligament insertion points as identified using
3DVIEWNIX. This ligament was represented by 1 element.
ta
59
x x
F
igure 11: ATFL identification plane, ligament insertion points and placement on hindfoot bones
Calcaneofibular Ligament (CFL)
Aligning a sagittal plane with the fibula and rotating it until it intersected the calcaneus
identified the CFL. The plane was then translated until the CFL could be identified as
shown in Figure 12. This ligament is located directly under the peroneal tendons, which
ere used as a reference. The ligament has a cylindrical structure that is discernible from
e surrounding structures and was therefore represented by 1 linear element.
w
th
60
x
x
Figure 12: CFL identification plane, ligament insertion points and placement on hindfoot bones
Posterior Talofibular Ligament (PTFL)
Aligning a coronal plane and translating it posterior to the talus identified the PTFL. This
ligament has a web-like structure that a the fibula[101] and was therefore
represented by 2 elements as shown in Figure 13.
sp ns out from
x xxx
Figure 13: PTFL identification plane, ligament insertion points and placement on hindfoot bones
61
Deltoid Ligament
Posterior Tibiotalar Ligament (PTTL)
Aligning a coronal plane and intersecting it with the posterior portion of the talus
identified the PTTL. This structure is thick and has deeper fibers that are laterally
oriented and superficial fibers that are vertically oriented[101]. Because of its thickness, 2
deeper elements and 2 supe icial elements were identified anteriorly and posteriorly as
represent this structure was conducted.
rf
shown in Figure 14. A parametric study on the numbers of structures that are necessary
to
Figure 14: PTTL identification plane, ligament insertion points and placement on hindfoot bones
Tibiocalcaneal Ligament (TCL)
A coronal plane intersectecting the sustentaculum tali identified the TCL. This structure
is difficult to discern from the TSL, which lies anterior to it, therefore one element
extending from the medial portion of the medial malleolus to the tip of the sustentaculum
tali represented it. As shown in Figure 15.
x xx
x
62
Figure 15: TCL identification plane
TibioSpring Ligament (TSL)
A coronal plan anterior to the s
ligament fans out from the anterio
spring ligament. 1 structure was se
2 additional structures anterior
parametric study on how different
Figure 16: TSL identification plane
x
, li
ust
r p
le
to
re
x
x
, lig
x
gament insertion points and placement on hindfoot bones
entaculum tali was used to identify the TSL. This
ortion of the medial tibia and inserts broadly into the
cted to represent this structure as shown in Figure 16.
the one shown were also selected to perform a
presentations of this ligament affect joint mechanics.
ament insertion points and placement on hindfoot bones
63
Anterior Tibiotalar Ligament (ATTL)
The ATTL was identified either by translating a sagittal plane medial-laterally across the
medial aspect of the tibia or by translating a coronal plane anterior-posteriorly. This
structure does not exist in all patients. In the specimens it was identified as the deeper
ligament structure spanning the ankle joint on the anterior medial side as shown in Figure
17.
Figure 17: ATTL identification plane, ligament insertion points and placement on hindfoot bones
Subtalar Ligaments
Interosseous Talocalcaneal Ligament (ITCL)
Aligning an oblique oriented plane perpendicular to the tarsal canal identified the ITCL
as shown in Figure 18. The plane was then moved from a posterior medial position to an
anterior lateral position 2 or more insertions were identified approximately every 3 mm.
ue to its long and thick structure, a total of 12 insertions and 11 insertions were marked
along its course for the in vivo and in vitro models, respectively.
x
x
D
64
x x
x x
Figure 18: ITCL identification plane, ligament insertion points and placement on hindfoot bones
Cervical Ligament (CL)
An obliquely oriented plane transecting the sinus tarsi identified the CL as shown in
Figure 19. The plane was translated until the borders of this structure were identified.
Eight insertion points along the periphery of the ligament were identified and used to
represent its geometry. Four of these points are shown below (Figure 19).
x
x
x
x
Figure 19: CL identification plane, ligament insertion points and placement on hindfoot bones
65
Ligament Mechanics
Each ligament was modeled as a tension-only element with non-linear load (T) –
strain ( )ε relationship as described by equation 14 below[102]. This expression was
derived using Fung’s quasi-linear viscoelastic theory[103]. The constants, A and B, were
btained in previous studies[102] by fitting the equation to experimental load-
llateral ligaments bone-ligament-bone preparations.
o
displacement tests for individual co
( ) ( ) .1,1),0, ),2,1((*))2,1(* 0.1 1( 00 ++−= LLMMDMSTEPMMVReAT Bε
ε Equation 14
he VR term monitored the magnitude of the first time derivative of the displacement
vector between the ligam
STEP functions. STEP func us pro co s transitions from the ON
and OFF states of parameters voi erical discontinuities, which may make the
system dynamics unsolvabl gen th P n (equation 15, below)
monitors the independent variable, , and activates with an initial value, , when
quals . The function then increases cubically until it reaches its final value, , when
T
ent insertion points, M1 and M2. The expression also includes a
tions are ed to vide ntinuou
. This a ds num
e. In eral, e STE functio
A 0h A
0x 1he
A equals x . 1
( )1100
The STEP function in expression 14 above monitored the magnitude of the distance
between ligament insertion poin
,,,, hxhxASTEP Equation 15
ts, M1 and M2. When the ligaments exceeded their
neutral position length, , a force between insertion points developed; otherwise the
ligament force remained zero.
0L
66
No structures were assigned time dependent relaxation properties. All loading times were
Previous studies characterized the non-linear Load-Strain properties of the collateral
ligaments[103]. The constants used to represent each structure in equation 14 are shown
in Table 3.
Table 3 o n consta s[102]
ament A B
at most 3 seconds and minimal relaxation occurs (< 10% force decrease) over this time
period[102]; therefore this component was not included in the model of the ligament
force properties.
Collateral Ligament Properties
: Ligament n n-linear load-strai equation nt
Lig R2
ATFL 7.18 12.5 0.965 CFL 0.20 49.63 0.828 PTFL 0.14 44.35 0.983 PTTL 1.34 28.65 0.999 TCL 0.51 45.99 0.543
ATTL 2.06 20.11 0.989
The tibiospring ligament’s nonlinear mechanical properties were not identified in the
material characterization study. This structure has a substantial stiffness[39], therefore,
like previous studies, the tibiospring and tibiocalacaneal ligaments may have been
lumped together during testing. Therefore, the force provided by these two structures was
divided equally between them.
67
Subtalar Joint Ligament Properties
The subtalar ligament’s structural properties have not been characterized, therefore their
load-strain properties were represented as a function of their calcaneal insertion areas as
calculated using point data obtained from the measure feature in 3DVIEWNIX (Table 4).
Since the ITCL and CL appear to have similar physical structures than the ATFL [67],
this ligament was scaled by a factor of the ratio: AreaATFLAreaITCL , where the ATFL area was
obtained in the literature[39].
Ligament In vivo
Table 4: Subtalar ligament insertion area
( )2mm In vitro ( )2mm ITC 54.92 90.69 LCL 20.7 8.06
Cartilage Mechanics
The force developed between contacting articular surfaces was defined as a non-linear
function of the penetration depth, x and the penetration velocity, , as shown in
quation 16 below. The penetration depth was scaled by a stiffness, , which was based
x&
kE
on the compressive modulus of cartilage[104]. The penetration term was also scaled by
an exponent term, e , which modeled the nonlinear compressive properties of
cartilage[105].
( ) ( )e xcdxSTEPxkForce &,,0,0, max+= Equation 16
68
The damping ratio, c , increased to its maximum value as a STEP function of the
penetration, x . When the penetration reached the value of ping reached its
ssigned va function controlled instantaneous changes in the damping
he stiffness value, , was derived using the experimentally determined compressive
odulus of carti
maxd , the dam
lue, c . The STEPa
force. The damping coefficient’s value was chosen to be 0.1 for the in vitro model and
1.0 for the in vivo model. Sensitivity analyses were performed to assess the affect of this
parameter on model mechanics. Finally, all contacts were frictionless to mimic cartilage’s
low coefficient of friction[103].
Derivation of Contact Stiffness
kT
lage, E . ( E =0.374 MPa) [104, 106, 107]. The modulum s was scaled by
the local average area, , of the polygons comprising eac sh at the
articulating surfaces, and thickness, , of the articular cartilage at each joint as shown in
equation 17 below.
A h bone surface me
t
tAEk = Equati
he local average area of one polygon comprising each polygonal bone surface was used
as the area scale factor. One polygon’s area was used because the RAPID interference
detection algorithm seeks to divide the geometry of each polygonal structure into the
smallest possible collection of polygons. Therefore, the contact force will be dependent
on a small area of contact. The coordinates of 5 sample triangles comprising the tibia
on 17
T
69
portion of the ankle joint articulating surface ere collected and the area of each triangle
was calculated. The avera
The average cartilage thickness (the gap between cortical bone layers of articulating
bones) was determined using the 3DVIEWNIX measure tool. Joint gaps were sampled at
the ankle joint tibio-talar surface and the subtalar joint posterior talo-calcaneal
articulation along a sagittal plane that was translated until it bisected the transmalleolar
axis. The thickness values are summarized in Table 5. Figure 20 below shows the
locations where cartilage thickness was measured using the 3DVIEWNIX Measure tool.
Cartilage thickness Contact Stiffness
w
ge area as 0 8 mw .8 m2.
Table 5: Intercortical distance at ankle joint and subtalar joint
mm mmN Joint Location
In vivo In vitro In vivo In vitro Anterior 2.40 1.60
Mid 2.70 2.30 Posterior 2.50 3.40 Ankle
Average 2.53 2.43
0.130 0.135
Subtalar Posterior 3.10 3.40 0.106 0.097 Talo-fibular Mid 3.50 2.50 0.094 0.132
70
x xxx xx
xx
Contact Penetration Exponent Derivation
The contact force is modeled as a non-linear function of the penetration distance as
previously shown in equation 16 above. The choice of exponent, e , was based on
cartilage’s non-linear behavior under axial loading[105]. Physiologically, the cartilage
cannot exceed a compressive axial strain of 100%. (i.e. the cartilage cannot compress
greater than its original thickness). Therefore, an exponent was chosen that would
generate very high compressive forces so that bone
Figure 20: Joint gap measurement points
penetration (i.e. cartilage
ompression) would not be greater than the average cartilage thickness (3 mm) at the
indfoot articulations. (Figure 21).
c
h
71
e = 7
e = 8
e = 9
e = 10
Figure 21: Contact force shown as a function of penetration distance ( ≤ 3 mm) for
107 ≤≤ e (k=0.116)
The exponent, , was chosen based on these data because, assuming a 3 mm cartilage
thickness, the contact force rose asymptotically, allowing no greater than 86%
compressive strain (2.6 mm penetration), (Figure 21).
FORCING FUNCTIONS AND BOUNDARY CONDITIONS Each hindfoot model was loaded to match the experiments performed with that subject.
In the model, the tibia and the fibula were fully constrained. The models were simulated
without gravitational forces. Cyclic and static loads (moments and forces) were applied to
the calcaneus about the joint axes defined by the International Society of
Biomechanics[108] for describing motion at the ankle joint complex. A torque producing
inversion / eversion at the ankle joint complex was applied about the e2 axis and a torque
e = 6
9=e
72
p
produc r each
simulation, the ankle joint complex was constrained to match the external constraints
imposed by the loading devices in the experiments (T
Table 6: The motion constra n th omp ivon ea
α = plantarflexion / dorsi ers γ = internal / external rotation, q2 = anterior / posterior drawer, q3 = compression / distraction
Motion
roducing internal rotation / external rotation was applied about the e3 axis. A force
ing anterior drawer was applied to the calcaneus along the e2 axis. Fo
able 6).
ints imposed omodels i
flexion, β = inv
e ankle joint cch movement
ion / eversion,
lex for the in v and in vitro
Model Loading M
Drawer Rotation ode
Inversion Anterior
Static α, γ, q2 α, β, γ -
Cyc α, γ α, β, γ α, β, q3lic , q2 in vivo
α α, β α Cyclic
Static α, γ, q2 α, β, γ - in
Cyclic α, γ, q2 α, β, γ α, β, q vitro
3
The in vitro and in vivo models were loaded to match the conditions of the experiments
able 7). The static loads were applied to the model to match the loading conditions in
the ankle loading device (ALD) used in the sMRI study. In the in vivo model, cyclic loads
were applied to match the loading conditions when the subject was tested in the Ankle
Flexibility Tester (AFT).
(T
73
Table 7: The loads applied for the in vivo and in vitro models and the corresponding rise time (for static loading) or period (for cyclic loading) of loading in each movement.
Inv=inversion, Ev=eversion, Int Rot=Internal Rotation, Ext Rot=External Rotation, Pflex=plantarflexion, Dflex=dorsiflexion
Motion
Model Dflex (+)
[Rise Time
(s)]
Loading
Mode
(Load
Device)
Inv (+) / Ev (-) {N-m}
[Rise Time, Period (s)]
Int Rot (+) / Ext Rot (-)
{N-m} [Rise Time, Period (s)]
Anterior Drawer (+)
{N} [Rise Time, Period (s)]
Pflex (-) /
{N-m}
or Period
Static (ALD) - + 2.26
[ 3 ]
+150
[ 3 ]
+ 7.5 / -7.5
[ 3 ]
Cyclic (ALD) [ 6 ] [ 6 ] [ 6 ] [ 12 ]
+ 3.4 / -3.4 + 3.4 / -3.4 +150 + 7.5 / -7.5 in vivo
Cyclic (AFT) + 2.6 / -2.6 + 2.9 / -2.9 +150
[ 6 ] -
[ 6 ] [ 6 ]
Static (ALD) - + 3.4
[ 3 ]
+150
[ 3 ]
+ 7.5 / -7.5
[ 3 ] in vitro
Cyclic (ALD) + 3.4 / -3.4
[ 6 ]
+ 3.4 / -3.4
[ 6 ]
+150
[ 6 ]
+ 7.5 / -7.5
[ 12 ]
74
MODEL MEASUREMENTS
Kinematics
joint rotation is described as finite helical axis rotation of the
coordinate system embedded in the calcaneus relative to the coordinate system embedded
in the talus. Centroidal translations are obtained directly from the information on the
location of the centroid in a coordinate system fixed to another bone at any position.
The finite displacements produced between the bones of the ankle complex (tibia
and talus, talus and calcaneus, and tibia and calcaneus) in response to applied loads were
computed and described using two motion description techniques: finite helical axis
rotations and Grood and Suntay parameters. These two techniques are complementary
since the first provides information on the net amount of rotation and translation while
the second decomposes the motion into six clinically relevant components, which can be
easily interpreted and correlated to the applied loads. The two techniques are briefly
described below.
Finite helical axis rotation and centroidal translation
Finite helical axis rotation is a well-established technique used to describe the
three-dimensional rotation of a rigid body in space[109]. Accordingly, a finite angular
displacement is described as a rotation by an amount Φ about an axis in a direction
defined by a unit vector n. The angular motion of a bone was described by using the
finite helical axes expressed in an inertial coordinate system attached to another bone. For
example, subtalar
75
The helical axis parameters and the centroidal translations of the ankle joint complex and
the ank
instant after loading. These data were then exported into a TXT file. A MATLABTM m-
file then calculated these motions (Appendix B). All parameters were calculated in the
ADAMS model and compared to the experimental data for the subject upon whom the
model was based.
le joint were defined relative to the centroidal reference of the tibia. Motion at the
subtalar joint was defined relative to the centroidal reference of the talus. The ADAMS
software automatically calculated the inertial references for each bone’s geometry. All of
the in vivo model’s centroidal bone references were oriented as follows: x had a posterior
direction, y had an inferior direction and z had a lateral direction. All of the in vitro
model’s centroidal bone references but z were oriented similarly: x had a posterior
direction, y had an inferior direction and z had a medial direction. The centroidal
reference axes defined by 3DViewnix were matched with those of ADAMS and assigned
the same axis name and direction to ensure that the experimental data and numerical data
were described in the same manner.
In order to calculate the helical axis parameters and centroidal translations, ADAMS
measured the following directional cosine matrices: the talus and calcaneus relative to the
tibia, and the calcaneus relative to the talus at the time instant before loading and the time
76
Grood and Suntay Parameters
Grood and Suntay[110] first introduced their anatomical joint coordinate system
to provide a clinically relevant motion description for the knee joint. The International
Society of Biomechanics[108] adopted their approach as a standard for describing the
motion of the ankle complex. In this technique, joint motion is described as rotation about
and translation along three axes. Two of these axes are embedded into the two bones
comprising the joint while the third axis is mutually perpendicular to the other two. In the
present study, three such anatomical coordinate systems were defined - one for the ankle
complex, one for the ankle joint, and one for the subtalar joint all shown in Figure 23
below. In order to describe the motion in terms of the Grood and Suntay parameters, the
anatomical axes defined in Figure 22 were identified in each hindfoot model. These were
obtained from the three points corresponding to the tip of the lateral malleolus - A1, the
medial malleolus - A2, and the centroid of the tibia on its proximal truncated end - A3.
An anteriorly oriented axis defined to intersect the bisection of the transmalleolar axis
and the medial-to-lateral midpoint of the talar head was defined as shown in Figure 22.
77
Figure 22: Definition of anatomical landmarks for Grood and Suntay parameters in ADAMS
•
•
•A1
A2
Sagittal Plane A3
78 78
For a right foot, the transmalle y, while for a left foot, it was
directed medially. The joint motion parameters were calculated using the directional
cosine matrices, , for each joint as shown in equations 19-21. The matrix was first
calculated for each joint (Equation 18). It represents the relationship between the
anatomical reference for lative to the tibia-fibula
complex at the ankle joint, the calcaneus relative to the talus at the subtalar joint, and the
calcaneus relative to the tibia-fibula complex at the ankle joint complex.
⎢
⎣
⎡••••••
10000
0
0
ZYkJjiJXkIjiI
Equation 18
The following relations between each joint’s direction cosine matrix and the Grood and
Suntay parameters were derived using vector algebra for the ankle joint complex:
olar axis was directed laterall
[ ]B
each joint as follows: the talus re
[ ]⎢⎢
•••=
kKjKiKB
⎥⎥⎥⎥
⎦
⎤
⎢
IJ
79
Plantarflexion / dorsiflexion
22
121tan B−α B
=
Inversion / Eversion
321cos
2B−−=
πβ
Internal rotation / External rotation
33B
311taB−=γ
Medial shift / Lateral shift
n
01 Zq =
Anterior drawer / Posterior drawer αα sincos YXq 002 −=
Compression / Distraction 3202201203 BZBYBXq ++=
for the ankle joint:
Plantarflexion / dorsiflexion
11
211tanBB
−= −α
Internal rotation / External rotation
311cos
2B−−=
πβ
Inversion / Eversion
33
321tanBB−=γ
Medial shift / Lateral shift
01 Zq =
Compression / Distraction αα cossin 002 YXq −−=
Equation 19
Equation 20
80
Anterior drawer / Posterior drawer
3102101103 BZBYBXq ++=
and for the subtalar joint:
Inversion / Eversion
22
321tanBB
−= −α
Plantarflexion / dorsiflexion
121cos
2B−−=
πβ
Internal rotation / External rotation
11
131tanBB
−= −γ
Anterior drawer / Posterior drawer 01 Xq =
Medial shift / Lateral shift
αα cossin 02 ZYq −−=
Compression / Distraction 3202201203 BZBYBXq ++=
where are the elements of the matrix Bij [ ]B .
These parameters were calculated in the ADAMS model and compared to the
experimental values for the subject upon whom the model was based.
Equation 21
81
k
j, e3J
An
Figure 23: Grood and Suntay joint axes definitions based on anatomical reference frames for the ankle joint complex, ankle joint and subtalar joint
e1
i, e1
e2
, e3
e2
K,
e2
j, e3
i
kle Joint plex
Ankle Joint Subtalar Joint
i k
K, e1
j
k
j
I
Com
I
J
ki
82
The signs to each joint’s direction of motion are summarized in Table 8 below.
Table 8: Directions for Grood and Suntay joint parameters at ankle joint complex (AJC), ankle joint (AJ) and subtalar joint (STJ) for right (R) and left (L) feet
l=lateral, at=anterior, po=posterior, I=inferior, s=superior
AJC AJ STJ
p=plantarflexion, d=dorsiflexion, iv=inversion, ev=eversion, it=internal protestation, et=external rotation, m=medial,
Parameter R L R L R L
α º
+ (p)
- (d)
+ (p)
- (d)
+ (p)
- (d)
+ (p)
- (d)
+ (ev)
- (iv)
+ (iv)
- (ev)
β º + (iv)
- (ev)
+ (ev)
- (iv)
+ (et)
- (it)
+ (it)
- (et)
+ (p)
- (d)
+ (p)
- (d)
γ º + (et)
- (it)
+ (it)
- (et)
+ (iv)
- (ev)
+ (ev)
- (iv)
+ (et)
- (it)
+ (it)
- (et)
q1 mm
+ (l)
- (m)
+ (m)
- (l)
+ (l)
- (m)
+ (m)
- (l)
+ (po)
- (at)
+ (po)
- (at)
q2 mm - (at) - (at) - (s) - (s) - (l)
+ (l)
- (m)
+ (po) + (po) + (i) + (i) + (m)
q3 mm
+ (i)
- (s)
+ (i)
- (s)
+ (po)
- (at)
+ (p)
- (a)
+ (i)
- (s)
+ (i)
- (s)
83
Joint Range of Motion and Flexibility
For the continuous load-displacement characteristics 3 cycles of data were collected. For
each test, the data from the second cycle was analyzed. The average load-displacement
characteristics (either moment relative to angular displacement or load relative to linear
displacement) were divided into two loading regions, 0-50 and 50-100% of maximum
load[60]. Flexibility was defined as follows: primary flexibility was the displacement
relative to the applied load in the direction of the applied load; coupled flexibility was
defined for the inversion-eversion test as the ratio between the amount of internal-
external rotation and the applied moment in inversion-eversion; early primary flexibility
was the ratio between the displacement and change in load for the 0-50% region; late
ary flexibility was the d the applied load for the
ary flexibility was the ratio between the displacement
tire 0-100% loading range (summarized in Figure 24),[60].
prim ratio between the displacement an
50-100% loading region; total prim
and the applied load for the en
Joint range of motion was defined as the maximum displacement attained in each loading
cycle. These parameters were calculated in the ADAMS model and compared to the
experimental values for the subject upon whom the model was based.
84
-50
-40
-30
-20
-10
0
10
20
40
-8 -6 -4 -2 0 2
Bet
a [D
egre
es]
30
50
4 6 8
INVERSION
EVERSION
Loading
Loading
∆Y1
∆X1
∆Y2
Early Flexibility =∆Y1 / ∆X1Late Flexibility =(∆Y2-∆Y1) / ∆X2Total Flexibility = ∆Y2 / (∆X2 + ∆X1)
∆X2
Torque [N-m]
Figure 24: Joint flexibility parameters
Ligament Strain and Force
For each simulation, the amount of ligament elongation, strain and force were measured.
Ligament strain was defined as the magnitude of the distance between ligament insertion
points divided by the initial ligament length after closing the inter-cortical gap at the
ankle joint and subtalar joint.
Contact Force Magnitude and Location
The magnitude of the contact forces at each articulating surface (Tibia-Talus, Fibula-
Talus, Talus-Calcaneus, Calcaneus-Fibula) ulation was recorded in each steady state sim
for all conditions. Preliminary contact location data was discussed for the ankle joint of
the in vitro model under inversion cyclic loading.
85
EXPERIMENTAL EVALUATION
Subjects
Data were collected from 1 healthy volunteer, age 48. An orthopaedic surgeon examined
the subject prior to testing and verified that no foot or ankle pathologies existed. Data
were also collected from an unembalmed cadaver leg disarticulated at the knee stored
ozen and thawed to room temperature before testing. fr
Experimental Tools
Ankle Flexibility Tester (AFT)
The AFT is a manually operated six degree-of-freedom instrumented linkage that
quantifies the load displacement characteristics of the hindfoot[111]. Forces and
moments are applied along clinically relevant axes (e1, e2 and e3 as shown in Figure 25
below), and the subsequent load-displacement data are recorded using a data acquisition
system. Previous studies showed that the device detects changes in the flexibility of the
hindfoot resulting from damage to the lateral ligaments[111]. The intraclass correlation
coefficient calculated from the test-retest data indicated a reliability higher than
0.85.[111] Previous evaluations of the device showed it had a positional accuracy of
better than 0.5 mm for translation and 1.2° for rotation.[111]
86
Figure 25: The Ankle Flexibility Tester
Following a decade of use, the AFT was in disrepair and unusable, therefore it was
completely overhauled. A test platform was constructed to mount both the AFT and a
chair so that the system could be used for in-vivo and in-vitro studies. The chair was
mounted on sliders that allowed it to be adjusted medially and laterally and anteriorly and
posteriorly to accommodate patient’s right and left feet and patients of different heights.
The AFT was mounted on a sliding platform so that it could be moved towards or away
from the patient’s foot.
The linear bearings allowing translation along the e1 and e2 axes were replaced with
heavy-duty models with higher loading tolerances. The e1 axis was fit with a locking
87
screw, which allowed the foot to be locked in 30° increments of plantarflexion or
dorsiflexion. A constant force spring was mounted parallel to the e2 axis between the AFT
ame and the foot mounting plate to offset the footplate’s weight. Therefore, the weight
of the footplate was not transferred to the hindfoot of the test subject. The linear
poten the
e3 axis was replaced with a pneumatic cylinder. This allowed an axial load to be applied
through the foot for simulating quiet standing loads. The base of the device was
reinforced so that it did not buckle when the patient’s foot was loaded. The electrical
wiring connecting the sensors of each axis to the data acquisition system was replaced to
eliminate noise in the data readings.
MR Compatible Ankle Loading Device (ALD)
The ALD, shown in Figure 26 below, is an MR compatible, non-metallic loading device
constructed to fit inside a 1.5 Tesla commercial MRI machine. This 6 degree-of-freedom
linkage allows unconstrained hindfoot motion and locks the joint in any position. The
device axes follow the Grood and Suntay joint coordinate system applied to the ankle
joint complex[108].
fr
tiometer measuring translation along e2 was also replaced. The linear portion of
88
Figure 26: The MRI compatible Ankle Loading Device
For in vivo testing, the patient lay prone on the MR gurney and the testers secured his/her
shank to the linkage. The foot was then placed on the footplate and the heel was secured
with medial, lateral and posterior clamps. For in vitro testing, the distal tibia and fibula
were cemented into a short 8 cm diameter plastic PVC cylinder and secured to one end of
the ALD. The calcaneus was rigidly fixed to the footplate with a threaded rod. A v-
shaped clamp immobilized the tibia for the anterior drawer test. A detachable u-shaped
aluminum bracket enabled the application of the inversion moment and the anterior
drawer force. The loads were applied outside the scanning room so that electrical strain
gauge torque and force sensors could be used to make the measurements without
influence from the strong magnetic field of the MR scanner. The external movement of
the hindfoot was measured between the footplate and the base through linear and angular
scales attached to the ALD.
89
Figure 27: Definition of anatomical landmarks for Grood and Suntay parameters in MRI
MR Scanner
A 1.5 Tesla commercial GE Signa MR Scanner was used. Two 3-inch single loop RF
coils and one 5-inch single-loop coil were placed on the two sides of the ankle and
underneath the heel. They were configured as a multi coil receiver. The scanning protocol
consisted of a 3D Fast Gradient Echo pulse sequence with a TR/TE/flip angle of 11.5
ms/2.4ms/60˚, a 512x256 in-plane acquisition matrix, a ±31.2 receiver bandwidth, and a
180 mm field of view. Sixty 2.1 mm-thick contiguous slices were collected.
Experimental Testing Procedure
In vivo Testing
The leg was placed in the ALD and aligned in a neutral position, as defined by the
International Society of Biomechanics[108]. The base of the device was rotated to align
the dorsiflexion/plantarflexion axis with the intermalleolar axis (Figure 27). The anterior
90
drawer bar was placed over the tibia. The foot was placed on the footplate and aligned
second ray was parallel to the centerline of the footplate. The
until it reached a value of 150N,
and the ALD was then locked in this loaded position. The table was rolled again into the
MR bore, and another MRI data set was collected.
For testing in the AFT, the subject was first secured in the device and aligned in the
neutral position similarly to that described above. The data acquisition system began
recording and cyclic moments were applied to the patient with an instrumented torque
handle starting in inversion / eversion, followed by internal rotation/ external rotation. An
instrumented force application handle was then used to test the subject in anterior drawer.
such that the base of the
heel was then clamped in place. The axes were locked, neutral position readings were
taken from each of the axes on the ALD, and an initial MR scan was acquired in this
neutral position.
After completing the neutral scan, the MRI scanner table with the subject was rolled out
of the scanner room away from the magnetic field. The ALD was unlocked and the
testers manually applied an inversion moment with an instrumented torque handle until it
reached a value of 2.26 Nm. The ALD was then locked in the loaded configuration, the
amount of ankle complex inversion was recorded directly on the ALD, the table was
rolled back into the scanner bore, and an MRI data set was acquired. Following the MRI
scan, the table was again moved out of the scanner room, and the inversion load was
removed. Then, by using an instrumented force application handle, the ankle complex
was loaded in anterior drawer. The load was increased
91
Data was recorded in each test for 6 cycles of approximately 6 seconds each to the loads
defined in Table 7 above. Data was recorded after each set of movement cycles were
completed.
In vitro Testing
After thawing the specimen to room temperature, it was aligned and fixed in the ALD in
the neutral position as described earlier. A threaded rod was secured through the
calcaneus and secured to the footplate. The in vitro MR testing procedure was exactly the
same as described above, with the exception that it was tested after the ATFL was
sectioning and after combined ATFL and CFL sectioning. The loads applied to the test
specimen followed those summarized in Table 7 above.
MODEL SIMULATIONS
Simulation Settings
All hindfoot simulations were solved with ADAMS’ default integrator, GSTIFF and the
Index 3 formulation for the equations of a mechanical system. The integrator error was
set to 0.001, the maximum integrator step size (hma s and the initial step
size (hinit) was 0.001 seconds. The default contact settings were used for the simulation.
These are use of the RAPID Geometry Library and a faceting tolerance of 300.
x) was 0.01 second
92
Preliminary Simulations
osition
do not include the cartilage geometry at the joint articulating
rfaces because each bone was segmented at the level of the cortical bone. As a result,
there were gaps of up to 3 mm between bones. In order to have contacting bone surfaces
for the model simulations, the ligament lengths were shortened to bring the bones from
the MRI neutral to a closed-gap neutral position. The goal of this procedure was to close
the bone gaps, with minimal bone orientation changes. The procedure was performed in
the following steps: 1. pretension all subtalar ligaments with 1 N load to close subtalar
joint gap; 2. record subtalar ligament lengths in joint-closed position; 3. adjust subtalar
ligament initial length in ligament force formulation; 4. apply 5 N axial load to calcaneus
along e3 axis of ankle joint complex to close ankle joint gap; 5. record new collateral
ligament lengths; 6. incorporate new collateral ligament lengths into their load-
displacement formulations. Each loading simulation began from the closed gap neutral
position (Figure 28) after these steps were completed.
Joint Neutral P
The hindfoot models
su
93
In vivo Model In vitro Model
neutral position Figure 28: Anterior view of the in vivo (Left) and in vitro (Right) hindfoot models in the closed-gap
94
Evaluation Studies
In Vivo Model
Steady-State Loading
Stress MRI was used to test the in vivo volunteer under inversion (2.26 N-m moment) and
anterior drawer (150 N force) loads. The loads were applied along the ankle joint
complex’s e2 axis defined by Grood and Suntay. The same constraints were applied to
the model as in the experiment. For inversion, ankle joint complex flexion and rotation
were constrained in neutral. For anterior drawer all rotations were constrained in neutral
across the ankle joint complex. The experimental helical axis data and centroidal
translations were calculated and then compared to those predicted by the numerical
model of the patient.
Cyclic Loading
The patient’s continuous load-displacement data (inversion/ eversion, internal rotation/
external rotation, anterior drawer) were measured in the AFT. The numerical model of
the patient was then loaded in the same manner as the experiments. For both the
experiment and the model inversion/ eversion and anterior drawer loads were applied
along the e2 axis of the ankle joint complex, while the moments were applied about e3 of
the ankle joint complex. The same constraints were applied to the model as in the
experiment (Table 7).
95
In Vitro Model
Steady-State Loading
The in vitro specimen was tested using sMRI under inversion (3.4 N-m moment) and
anterior drawer (150 N force) loads (Table 7). The experimental helical axis data and
centroidal translations were compared to those predicted by the numerical model under
the same loading conditions. The in vitro specimen was then tested after the ATFL was
sectioned and after both the ATFL and CFL were sectioned under inversion and anterior
drawer loads. The same constraints were applied to the model as in the experiment. For
inversion, ankle joint complex flexion and rotation were constrained. For anterior drawer
ll rotations were constrained across the ankle joint complex. The same experimental
calculated for these conditions and compared to those predicted by
a
measurements were
the numerical model under the same loading conditions.
Parametric Studies (Sensitivity Analysis)
Ligament Orientation
The calcaneal insertion of the CFL was translated in 5 mm increments anteriorly and
posteriorly relative to its original calcaneal insertion in the in vitro model in order to
determine the sensitivity of the predicted hindfoot kinematics to ligament orientation and
insertion location. A 3.4 N-m inversion moment was applied to the calcaneus. The torque
as oriented along the e2 axis defined by the Grood and Suntay definition for motion at
the ankle joint complex. The kinematics of the ankle joint complex were recorded and
compared for each CFL position.
w
96
Ligament Representation
Ligaments that have a broader area of insertion such as the medial collateral structures of
e deltoid ligament were investigated to determine the sensitivity of model kinematics to
used to represent them. The deep PTTL was represented as
with 2 to 4 structures spanning the anterior colliculus of the tibia to points on and anterior
to the sustentaculum tali of the calcaneus. When more than one element represented any
structure, their force contribution was divided amongst the structures by the number of
structures used to represent each ligament. The sensitivity of ankle joint complex, ankle
joint and subtalar joint kinematics in eversion (3.4 N-m moment applied to hindfoot)
were compared for each TSL/TCL representation.
Linear vs. Non-linear Ligament Mechanics
The contribution of non-linear ligament stiffness to the non-linear load displacement
characteristics of the hindfoot was determined by comparing the in vitro model with only
linear ligament stiffness properties to the same model using non-linear ligament load-
strain properties. The hindfoot was simulated in inversion-eversion cyclic loading (± 3.4
N-m) with a period of 6 seconds. The load-displacement graph was plotted and its
characteristics (path of loading-unloading, early and late flexibility) were qualitatively
compared to its non-linear ligament stiffness counterpart.
th
the number of force elements
1 to 3 elements spanning the intercollicular fossa of the tibia and the medial talar
tuberosity. The in vitro model was simulated with a 3.4 N-m inversion load and with the
ATFL and CFL removed. This condition was used because preliminary simulations
indicated that this structure experienced large elongation. The TSL/TCL was represented
97
Contact Damping
nges in the contact damping term at each joint were
explored by varying this term by levels above (1 Ns/mm) and below (.05 Ns/mm) its
chosen value (0.1 Ns/mm) in the in vitro model. All joint contacts (tibia-talus, talus-
calcaneus, talus-fibula, calcaneus-fibula) were assigned the same damping value for each
test. The hindfoot was simulated in inversion/ eversion cyclic loading (± 3.4 N-m) with a
period of 6 seconds. The inversion-eversion load-displacement graphs were plotted and
their characteristics (path of loading-unloading, early and late flexibility) were
qualitatively compared.
P
The sensitivity of model output to cha
rediction Studies
In Vivo ATFL Tear and Combined ATFL/CFL Tear
The in vivo model was tested under the steady-state loads and constraints described in
Tables 6 and 7 for the ATFL removed and combined ATFL+ CFL removed conditions.
ADAMS measured the joint contact forces, ligam
joint mechanics in terms of the Grood and Sunta rs de earlier. Where
possible the data were compared to data fro ratu plantarflexion
dorsiflexion, a 7.5 N-m cyclic mo applie 2 se eriod to match
previous experiments[112].
The in vivo model was also tested under the cyclic loads described in Table 7. For the
intact condition ADAMS simulated plantarflexion/ dorsiflexion (
ent elongation, strain and forces and
y paramete fined
m the lite re. In
ment was d over a 1 cond p
α ), inversion/ eversion
( β ), internal rotation/ external rotation (γ ) and anterior drawer (q2). ADAMS was used
to simulate all motions with the exception of flexion after the same ligaments described
98
above were removed. The flexibility data were measured in each condition for all loading
modes and compared to previous experiments
In Vitro Model Predictions
The in vitro model was tested in the same movements, described for the in vivo test. The
loading conditions and boundary constraints are summarized in Tables 6 and 7. The
model was tested both cyclically and under steady state loads with and without ligaments
removed. These data were then compared to existing data from the literature.
99
CHAPTER 4. RESULTS
BONE VOLUME COMPARISON
With the exce e in vivo cal eus m el, th ne v decreased as a
result of the surface refinement G e
change occurred when processing the in odel of the fibula (6.05% volume
decrease) (Tab e e g the in vivo
model was 3.5 e a vo change for the bones comprising the in vitro
structure was 3.6
Table 9: Volume comparison of bone geometries ed from 3Dview x and after applying the reduce nd ate s in AG TM
e
ption of th can od e bo
MA
olum
IC
es all
features used in GEO TM. The largest volum
vivo m
le 6). Th averag volume change for the bones comprisin
5%. Th verage lume
2%.
obtain ni noise a decim feature GEOM IC
Volum ( )V mm3
Su ect B 3DVIEWN
O CT
processing bj one IX GE
AfterMAGI M V∆%
Tib 58,808.25 5 6 6,177.1 -4.47 Fib 12,962.91 1 5 2 4,178. -6.05 Tal 34,702.92 32,878.88 -5.26 In vivo
36 70,008.69 1.57 Cal 68,928. Tib 4 .95 44,404.35 -3.00 5 7,77 F 11 -4.38 ib ,853.68 11,334.04 Tal 28,634.65 2 67,570.3 -3.72 In vitro
19 66458.65 -3.46 Cal 68,842.
100
SIMULATION TIMES
Each hindfoot model, running on a 2.6 GHz, Dell Optiplex GX270 with an Intel Pentium
4 microprocessor and 1.0 Gbytes of RAM, required 8 minutes to perform preprocessing.
Preprocessing includes calculating the jacobian matrix for integrating the equations of
otion and calculating the OBB boxes for the contact function. The OBB information
times were shorter (2 minutes). The statically loaded models’ simulation time after
preprocessing were approximately 7 minutes. The cyclically loaded models required an
additional 12 minutes to complete.
PRELI Y S ATIO
Hindfoot Neutral Position
m
was calculated once for each simulation session, therefore subsequent preprocessing
MINAR IMUL NS
The subtalar ligaments’ lengths decreased in order to close the inter-cortical gap at the
subtalar joint for both the in vivo and in vitro hindfoot models (Table 10). The more
vertically oriented collateral ligaments such as the TSL and TCL shortened when the
inter-cortical gap at the ankle joint was closed. The more horizontally oriented structures
such as the ATFL, the PTFL and the deeper portion of the PTTL shortened to a lesser
extent in both models.
101
Table 10: Comparison of ligament lengths and residual ligament forces in MR neutral and closed gap
mm Neutral mm
Remaining Tension
N
neutral positions for in vivo and in vitro models
Lo MR neutral Lo Closed-gap
Ligament
In vivo In vitro
In vivo
In vitro
In vivo
In vitro
Collateral Ligaments ATFL 11.0 5.03 11.08 4.55 0.32 0.45 CFL 15.49 15.49 10.05 14.60 0.03 0
PTFL1 12.57 21.62 13.08 19.58 0.05 0.73 PTFL2 18.91 16.81 19.56 15.30 0.05 0 ATTL 6.35 4.37 3.2 4.00 0.01 0 TSL 29.42 26.98 25.00 0 22.70 0 TCL 24.64 18.57 2 3 0.42 0.60 13.8 0.05
PTTL1 26 .5 .0 5.42 0.2 23. 4 4 5 0 5 1.1 PTTL2 .46 . .7 12.9 0 11 12 21 9 9 0
Subt iga alar L mentsIT cCL stru tures
1 4.06 5.51 2.42 4.43 0 0 2 5.63 7.4 . 6 0 7 4 38 .50 03 5.31 8.95 4.106 7.21 0 0.32 4 5.21 8.4 . 7.20 0 6 3 83 0 5 5.52 10.40 4.037 7.90 0 0 6 5.22 .0 .6 13.90 0 0 15 0 3 74 7 6.22 11.97 4.5 9.24 0 .98 75 0 8 5.87 .8 .5 .02 12 6 4 53 11.24 0.23 0 9 9.55 .3 .9 11.8 14 8 7 47 0.36 0 10 6.99 .3 .6 8.90 0 12 19 9 5 22 0. 11 8.93 18.75 7.722 12.10 0 0 12 6.75 - 5.716 - 0 -
CL structures 1 9.04 15.52 7.92 12.29 0.05 0.44 2 8.22 12.60 8.34 9.61 0 0 3 6.26 13.21 8.27 10.66 0 0 4 9.41 15.78 7.7 13.00 0 0
Closing the inter-cortical gap caused small changes in the orientation of the talus and
calcaneus in both the in vivo and in vitro hindfoot models (Table 11). All rotations were
less than 5.98º with the exception of external rotation at the subtalar joint (7.18º). The
102
talus translated superiorly relative to the tibia (q2 AJ = 2.76 mm in vivo, 1.49 mm in vitro)
indicating that the gap at the ankle joint closed. The calcaneus also translated superiorly
relative to the talus (q3 STJ = 1.40 mm in vivo, 1.69 mm in vitro ) indicating that the
subtalar joint space was closed.
Table 11: Kinematic changes at hindfoot resulting from closing inter-cortical gaps at ankle joint and
subtalar joints for In vivo and In vitro models
Ankle Joint Complex Ankle Joint Subtalar Joint Parameter
In vivo In vitro In vivo In vitro In vivo In vitro
α º 5.32 -0.17 1.05 4.27 -2.73 -2.65 β º -0.02 -1.76 1.09 3.07 3.43 -3.82 γ º -5.15 -0.65 -0.90 -2.96 -7.18 -3.86
q1 mm -0.43 2.88 0.33 -1.29 -1.23 0.36 q mm -1.59 1.35 -2.76 -1.49 0.69 -3.79 2 q3 mm -4.13 -0.66 -0.44 -0.01 -1.40 -1.69
EVALUATION STUDIES
In vivo Model Evaluation
Static Loading – Kinematics (sMRI Comparison)
a vector that is oriented about an anteriorly directed axis (x = -0.989). In the experiment,
rotation at the ankle joint complex occurred about a superiorly directed axis (y = -0.983).
At each joint, the numerically predicted unit vector about which rotation occurred
differed greatly from that of the experiment. The experimental centroidal translations
Under an inversion moment the numerically predicted screw axis rotation and centroidal
translation magnitude were greater than those measured experimentally at each joint
(Table 12). The model also predicted that rotation at the ankle joint complex occurs about
103
were all less than 1 mm at the ankle joint and the subtalar joint, and the model predicted
rger translations than th es of the model and the
experiment appear to b tiv ilar positions in response to the inversion
moment (Figure 29).
Table 12: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y neutra nversi the model and the test subject
Ankle Joint Complex Subtalar Joint
la is at each of these joints. The bon
e in qualita ely sim
, z) from l to i on for in vivo
Ankle Joint Parameter
Model Expt Diff Model Expt Diff Model Expt Diff Angle ° 7.65 2.27 -5.38 6.72 2.18 -4.54 10.17 1.74 -8.43
Rot Vec x -0.989 0.019 -0.418 0.698-0.112
-0.723
y -0.07 0.914 -0.811 -0.983
-0.656
-0.492
z 0.127
80.73º
0.405 0.216
121.72º
0.409 -0.52
96.05º
-0.145
Trans Mag mm 1.75 1.95 0.20 2.29 0.27 -2.02 3.77 0.88 -2.89
X mm 0 -1.59 -1.59 1 -0.16 -1.16 -1.1 -0.76 0.34 Y mm 1.7 1.13 -0.57 -1.2 -0.22 0.98 2.9 0.39 -2.51 Z mm -0.43 0.12 0.55 1.67 -0.02 -1.69 -2.14 0.2 2.34
104
Figure 29:Comparison of inverted in vivo model (L) and corresponding MR image data (R) for the intact condition
In terms of the Grood and Suntay parameters, the inversion moment caused the calcaneus
to invert with respect to the tibia (
β AJC=8.29º), (Table 13). Inversion of the ankle joint
complex was coupled with external rotation of the talus with respect to the tibia ( β AJ =
5.54º) and internal rotation (γ STJ = -6.20º) of the calcaneus with respect to the talus. No
rotation about the plantarflexion / dorsiflexion axis (α ) or the internal / external rotation
axis (γ ) occurred across the ankle joint complex due to the parallel constraint applied
between the tibia and calcaneus.
105
between neutral and inverted positions
Grood and Suntay
Parameters
Ankle Joint
Complex
Ankle Subtalar
Table 13: Grood and Suntay Parameters describing changes in vivo hindfoot model kinematics in
Joint Joint
α º 0 -2.33 -8.4 β º 8.29 5.54 -1.68 γ º 0 0.40 -6.20
q1 mm 0.55 -0.94 3.12 q2 mm 0.7 -1.5 -2.34 q3 mm 0.7 1.59 2.61
For the anterior test, the predic is r the ankle
joint and the subtalar joint differed fr the experime easured data by less than
0.45º (Table 14). At ch joint, the num ically predicted unit vector about which rotation
occurred differed greatly from that of the experime the ex nt, rotation
occurred at the ankle joint complex about a laterally directed axis. (z = 0.892) The
predicted centroidal translation magnitude was greater at each joint than the
odel and the
model appear to have moved more anteriorly than those of the test subject in response to
drawer numerically ted screw ax otation of
om ntally m
ea er
nt. In perime
experimentally measured translation by a factor of 4. In both the m
experiments, translation occurred primarily along an anteriorly directed vector. This
corresponds to the direction that the anterior drawer force was applied. The bones of the
the anterior drawer force (Figure 30).
106
Table 14: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal
subject
Ankle Joint Complex Ankle Joint Subtalar Joint
translations (Trans Mag, x, y, z) from neutral to anterior drawer for the in vivo model and the test
Parameter
Model Expt Diff Model Expt Diff Model Expt Diff Angle ° 0 2.36 2.36 3.89 3.44 -0.45 4.42 4.10 -0.32
Rot Vec x -0.240 0.219 0.338 -0.872 -0.754 0.835
y -0.959 -0.395 -0.286 -
0.487 -0.243 0.545
z -0.150 0.892 0.897 0.041 -0.610 -0.06
78.90º
96.82º
9
136.05º
TMag mm 8.11 2.17 -5.9 -3.87 4.08 0.62 -3.46 rans 4 4.31 0.44
X mm -7.70 -1.74 5.96 -0.42 3 60 -0.51 3.09 -4.07 .65 -3.Y mm 1.0 0.13 -0.49 0 0.022 -1.28 0 1.13 -0.90 0.41 1.3Z mm 2 -2 .0 0 .36 0.117 .24 1.09 0 4 -1.05 1.4 0.352 -1.05
Figure 30:Comparison of in vivo model (L) and corresponding MR image data (R) for in anterior
drawer
107
In terms of the Grood and Suntay parameters, no rotation occurred across the ankle joint
complex due to the rotational constraints applied between the tibia and calcaneus (Table
15). Substantial translation occurred along the axis that the anterior drawer load was
applied (q2 AJC=-7.76mm). The anterior translation occurred both at the ankle joint (q3
AJ=-4.09mm) and the subtalar joint (q1 STJ=-3.72mm). The model also predicted talar
dorsiflexion at the ankle joint (α AJC=-6.62º).
Table 15: Grood and Suntay Parameters describing changes in bone position from neutral to
anterior drawer positions as predicted by in vivo hindfoot model
Grood and Suntay
Parameters
Ankle Joint Complex Ankle Joint Subtalar
Joint
α º 0 -6.62 -2.24 β º 0 -1.02 3.33 γ º 0 -0.99 0.10
q1 mm 2.51 1.18 -3.72 q2 mm -7.76 -0.14 -1.47 q3 mm -0.28 -4.09 0.72
Cyclic Loading (Comparison to AFT Load-Displacement Data)
The patient’s ankle joint complex range of motion [ROM] in inversion, internal rotation
and anterior drawer was much greater than that predicted by the model. (Table 16) The
experimental range of motion in eversion and external rotation nearly matched the model
results. In the motions of inversion and anterior drawer, the numerically predicted early
flexibility nearly matched the experimental data of the patient (Table 16). In each
irection, the model exhibited higher early flexibility and much lower late flexibility
(Figures 32-34). The exp tial decrease in the late
d
erimental results did not show a substan
108
flexibility compared to early flexibility as shown Table 16 and displayed in Figures 31-
33. Rotation was coupled with inversion and eversion in both the model and the
experimental data as shown in Table 16. The model predicted that internal rotation was
coupled with inversion and external rotation was coupled with eversion, which was the
opposite of the experimental data.
Table 16: Comparison of early, late and total flexibility and range of motion [ROM] characteristics
of the in vivo model and the patient upon which the model was based. Coupled G = internal rotation [ - ] / external rotation [ + ] coupled with inversion / eversion
q2 = anterior drawer
Flexibility Early Late Total ROM [deg] [deg/N-m] [deg/N-m] [deg/N-m] Model Exp Model Exp Model Exp Model Exp
Inv 10.19 11.07 0.80 13.88 5.5 12.46 7.53 34.86 Coupled G -2.90 3.65 -0.18 2.82 -1.54 3.24 -4.08 9.06
Ev 14.62 7.75 1.81 8.57 8.21 8.16 18.80 13.10 Coupled G -4.21 2.32 0.42 3.05 -1.9 2.69 -5.02 6.64
Int 15.65 10.83 0.57 10.49 8.11 10.66 -15.50 30.17 Ext 16.16 9.10 0.76 6.84 8.46 7.97 13.90 18.50
q2 [mm/N] 0.084 0.141 0.007 0.142 0.046 0.142 7.76 24.51
109
-20
-10
10
30
T e
β [d
eg]
Figur 1: In v (thin lines)
inversion [ + ] / eversion [ - ]
20
40
0-3 -2 -1 0 1 2 3
orqu [N-m]
e 3 ivo experimental and numerical (thick lines) load-displacement curves for
-40
-30
-20
-10
0
10
20
-3 0 1 3
Torque [N-m]
γ [d
eg]
Figure 32: In vivo experimental (thin lines) and numerical (thick lines) load-displacement curves for
internal rotation [ - ] / external rotation [ - ]
Loading Inversion
Eversion
-2 -1 2
Loading
External Rotation Loading
Internatio
din
l Rota n Loa g
110
0
5
10
15
20
25
-5 45 95 145Force [N]
q 2 [m
m]
Anterior Drawer Loading
Figure 33: In vivo experimental (thin lines) and numerical (thick lines) load-displacement curves for anterior drawer [ + ]
vitro Model EvaluationIn
tatic Loading – Kinematics (sMRI Comparison)
The screw axis rotation at the ankle joint complex predicted by the in vitro model nearly
matched that of the experimental data (16.6% difference) under an inversion moment
ankle joint were much smaller than the
experimental values. At the subtalar joint, the model predicted twice as much rotation
than what was measured in the experiment. At each joint, the numerically predicted unit
vector about which rotation occurred differed greatly from that of the experiment. For
example, rotation at the ankle joint complex occurred about an axis that was directed
primarily in a posterior direction (x = 0.990), while the experimental rotation occurred
about an axis that was oriented laterally (z = -0.882). The calcaneus translated medially
S
(Table 17). The predicted rotations at the
111
r
The model also predicted much larger translations at the subtalar joint (4.93 mm) than
what was ex measu mm). The calcaneus appears to be more
internally rotated in the experiment than what was predicted in the model (Figure 34).
Table 17: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, ztranslations (Trans Mag, x, y, z) from ne inversion for tro model and the test subject
Ankle Joint Complex Ankle Joint Subtalar Joint
elative to the tibia in both the experiment (z = 7.86 mm) and the model (z=4.59 mm).
perimentally red (1.30
) and centroidal utral to the in vi
Parameter Model Expt Diff Model Expt Diff Model Expt Diff
Angle ° 13.41 11.5 -1.91 4.99 15.2 10.21 12.97 6.28 -6.69
Rot Vec x 0.990 0.449 0.395 0.644 0.986 -0.795
y -0.051 -0.144 0.018 -0.388 0.148 0.598 z -0.134 -0.882
55.25º
0.919 -0.659
110.99º
0.08 -0.098
134.68º
Trans Mag mm 5.07 9.05 3.98 2.02 2.39 0.37 4.93 1.30 -3.63
X mm -1.7 4.11 5.81 -0.26 -1.6 -1.34 -1.4 -1.10 0.30 Y mm -1.3 1.83 3.13 -2.00 1.36 3.36 0.70 0.62 -0.08 Z mm 4.59 7.86 3.27 -0.09 1.13 1.22 4.67 -0.28 -4.95
112
F om o ed model (L) and corresponding MR image data (R) for the in d
In terms of the Grood and Suntay parameters, the calcaneus inverted relative to the tibia
oin vitrig Cure 34: parison f inverttact con ition
( β AJC = -12.47º) and this motion occurred primarily at the subtalar joint (α STJ = 13.66º),
(Table 18).
113
Table 18: Grood and Suntay Parameters describing changes of in vitro hindfoot model kinematics from neutral to inverted positions
Grood and
Suntay Parameters
Ankle Joint Complex Ankle Joint Subtalar
Joint
α º 0 -2.12 13.66 β º -12.47 -0.51 -1.87 γ º 0 1.13 0.99
q1 mm -3.55 -0.15 1.52 q2 mm -0.17 -2.18 4.2 q3 mm -0.47 -0.35 0.63
Under anterior drawer loading, the numerically predicted centroidal translation
magnitude at the ankle joint complex and the ankle joint agreed with the respective
experimentally measured translation magnitude (26.6% difference at ankle joint complex
and 3.8% difference at ankle joint), (Table 19). At the ankle joint complex, the model
predicted that the translation would occur primarily along an anteriorly directed axis
t of rotation at the ankle joint
and the subtalar joint agreed with their experimentally measured values (13.8%
difference at the ankle joint and 9.6% difference at the subtalar joint). At each joint, the
numerical and experimental orientations of the unit vector along the helical axis differed
greatly. For example, rotation at the ankle joint complex occurred about an axis that was
directed primarily in a superior direction (x= -0.873), while the experimental rotation
while in the experiment translation of the calcaneus relative to the tibia was divided
almost equally along the centroidal axes of the tibia. The model overestimated the
magnitude of subtalar joint translation by 94.1%. In the model, no rotation occurred at the
ankle joint complex due to the rotational constraints imposed on the model, but rotation
did occur in the experiment (8.72º). The predicted amoun
114
occurred about an axis that was oriented laterally (y = -0.843). The pictures of the model
and the experiment in the loaded position are qualitatively similar (Fig
Table 19: Screw axis rotations (Angle), ical axis orientat Vec x, y, centroidal translations (Trans Mag, x, y, z) comparin anges from neutr anterior dr r the in vitro
mod the test subjec
Ankle Joint Complex nkle Join Subtalar Joint
ure 35).
hel ion (Rot z) and g ch al to awer fo
el and t
A t Parameter
Model Expt Diff Mode Expt Diff Model Expt Diff lAngle ° 0 8.72 8.72 7.75 6.81 -0.94 7.90 7.21 -0.69
Rot Vec x -0.873 0.005 -0.082 0.746 0.491 0.816 - -
y 0.218 0.843 -0.129 0.662 0.583 0.578 - - -
z 0.436 0.538 0.988 0.070 -0.648 0.003 86.84º 84.64º 137.70º
Trans Mag mm 6.70 5.29 -1.41 2.52 2.62 0.10 4.29 2.21 -2.08
X mm -6.30 -2.94 3.36 -2.13 -2.48 -0.35 -4.1 -1.64 2.46 Y mm -1.00 3.52 4.52 -1.1 0.42 1.52 0.10 1.34 1.24 Z mm -2.04 2.64 4.68 -0.78 -0.76 0.02 -1.252 -0.63 0.62
115
Figure 35:Comparison of in vitro model (L) and corresponding MR image data (R) in anterior drawer for the intact condition
In terms of the Grood and Suntay parameters, the model predicted that at the ankle joint
complex, translation occurred primarily in the direction of the anterior drawer load (q2 AJC
= -6.58 mm), (Table 20). The calcaneus also translated with respect to the talus (q1 AJC = -
4.61 mm). The translations occurring at the ankle joint complex were also coupled with
dorsiflexion of the talus at the ankle joint (α AJ = - 7.22º).
116
Table 20: Grood and Suntay Parameters describing changes In vivo hindfoot model kinematics from
Ankle Joint Subtalar
neutral to anterior drawer
Parameter Complex Ankle Joint Joint α º 0 -7.23 3.81 β º 0 -1.19 6.33 γ º 0 1.7 3.03
q1 mm 0.42 -0.01 -4.61 q2 mm -6.58 -0.12 2.2 q3 mm -1.18 6.47 -1.53
After removing the ATFL from the m el, the predicted screw axis rotation at the ankle
joint complex resulting from an inversion moment was in agreement with that of the
experimental data (-25.9% difference). The rotations pr by the m at the ankle
joint were much (67.2%) th e easured experimentally. The subtalar joint
rotations predicted by the model differed by 39.7% from the experimental measurement
od
edicted odel
smaller an thos m
of the cadaver in the same condition. The helical axis orientations calculated by the
model and in the experiment differed by over 90º at the ankle joint and the subtalar joint
and by less than 45º at the ankle joint complex. The model predicted much less
translation at the ankle joint complex than what was measured in the experiment (69.7%).
In both cases, the calcaneus translated medially relative to the tibia under an inversion
moment (z = 2.73 mm for the model and z = 10.00 mm in the experiment). The hindfoot
model appears to be more dorsiflexed than in the experiment (Figure 36).
117
Table 21: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal
model and the test subject
Ankle Joint Complex Ankle Joint Subtalar Joint
translations (Trans Mag, x, y, z) from neutral to inversion with the ATFL sectioned for the in vitro
Parameter
Model Expt Diff Model Expt Diff Model Expt Diff Angle ° 12.00 16.20 4.20 6.05 18.46 12.41 11.89 8.51 -3.38
Rot Vec x 0.990 0.620 0.394 0.580 0.975 0.115
y -0.050 -0.292 -0.067 -
0.677 0.220 -0.992
z -0.132 -0.728
0.917 -
0.450
97.98º
-0.002 -0.057
96.08º
43.57º
Trans Mag mm 3.50 11.54 8.04 2.33 1.44 -0.89 3.07 1.40 -1.67
X mm -1.10 5.66 4.56 -0.30 -1.02 -1.32 -0.70 1.08 1.78 Y mm -1.90 1.00 2.90 -2.30 0.92 3.22 0.40 0.29 -0.11 Z mm 2.73 10.00 19 2.96 -0.84 2.12 12.73 -0.24 0.43 0.
Figure 36:Comparison of inverted in vitro model (L) and corresponding MR image data (R) with the ATFL sectioned
118
In terms of the Grood and Suntay parameters, the calcaneus inverted relative to the tibia
( β AJC = -12.20º) and this motion occurred primarily at the subtalar joint (α STJ = 11.8º),
(Table 22).
Table 22: Grood and Suntay Parameters describing changes in in vitro hindfoot model kinematics from neutral to inversion with the ATFL sectioned
Parameter Ankle Joint Complex Ankle Joint Subtalar
Joint α º 0 -2.44 11.8 β º -12.2 0.33 -1.6 γ º 0 -0.53 0.19
q1 mm -3.95 0.01 2.2 q2 mm -0.18 -1.74 4.77 q3 mm -0.51 -0.83 0.17
he magnitude of the centroidal translation predicted by the model agreed with the
xperimental measurements at the ankle joint complex (6.8% difference) and the subtalar
tion was much greater at the ankle joint than at the subtalar
joint (7.47 mm at the ankle joint and 2.84 mm at the subtalar joint). The translation
predicted by the model was also much greater at the ankle joint than the amount
measured in the experiment (91.5% more translation predicted by model). Substantial
rotation occurred at the ankle joint in both the numerical model (11.65º) and in the
T
e
joint (15.4% difference), (Table 23). The model predicted that anterior translation of the
calcaneus at the ankle joint complex occurred primarily along the anteriorly oriented
component of the tibial inertial reference frame (x=-8.10 mm). In the experiment, the
calcaneus moved anteriorly (x = -4.88 mm) and medially (z = 6.43 mm) relative to the
tibia. In the model, transla
119
experiment (7.33º). This rotation occurred about an axis that was directed inferiorly in the
model (y = 0.930) x = 0.996) in the
experiment. The orientation of the helical axis was m in t cal model
and the experimental measurement differing by 66.42º at the ankle joint and up to 120.63º
at the ankle joint co plex. The pictures of the model experi the loaded
position are qualitatively similar (Figure 37).
Table 23: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to anterior drawer with the ATFL sectioned for the in
and about an axis that as directed poseriorly (w
uch different he numeri
m and the ment in
vitro model and the test subject
Ankle Joint Complex Ankle Joint Subtalar Joint Parameter
Model Expt Diff Model Expt Diff Model Expt Diff Angle ° 0 4.56 4.56 11.65 7.33 -4.32 12.32 3.72 -8.60
Rot Vec x -0.873 0.862 0.368 0.996 -0.029 0.998y 0.218 0.143 0.930 0.036 -0.835 0.028
z 0.436 0.486
120.63º
0.0002 0.086
66.42º
-0.550 0.051
91.39º - -
Trans Mag mm 8.20 8.80 0.60 7.47 3.90 -3.57 2.84 2.46 -0.38
X mm -8.10 -4.88 3.22 -7.05 -2.14 4.91 -1.00 1.87 2.87 Y mm -1.30 3.51 4.81 -0.10 1.45 1.55 -1.10 0.52 1.62 Z mm 0.07 6.43 6.36 2.48 2.93 0.45 -2.42 -1.51 0.91
120
Figure 37: Compari vit del co nd R image data (R) for anterior d it ion
In ter
com oth in the anterior direction (q = -7.82 mm) and inferiorly (q C = -8.28
mm le 2 e r tio r m t le (q -
7.0 Th l e su ial na on t el
son of the in ro o mrawer w
(L) and rrespo ing Mh the ATFL sect ed
ms of the Grood and Suntay parameters, translation occurred at the ankle joint
plex b 2 AJC 3 AJ
), (Tab 4). Th anterio transla n occu red pri arily a the ank joint 3 AJ =
2 mm). e mode also pr dicted bstant inter l rotati of the alus r ative to
the tibia ( β AJC =11.53º).
121
Table 24: Grood and Suntay Parameters describing changes in in vitro hindfoot model kinematics
from neutral to anterior drawer with the ATFL sectioned
Parameter Ankle Joint Complex Ankle Joint Subtalar
Joint α º 0 -1.49 -1.99 β º 0 11.53 4.69 γ º 0 -2.28 -10.63
q1 mm 3.07 0.29 -0.37 q2 mm -7.82 2.08 0.02 q3 mm 8.28 -7.02 -2.08
After the ATFL and the CFL were sectioned, an inversion moment increased the ankle
complex rotation from previous conditions where the CFL was intact (Table 25). Both the
model and the experimental rotation at the ankle joint complex were large (36.31º for the
odel and 25.46º for the experim diction differed by 42.7% from
ankle joint (31.53º) about an axis that was oriented both in the posterior direction (x =
0.743) and the lateral direction (z = -0.627). The rotation measured in the experiment at
e subtalar joint (7.55º) was much less than at the ankle joint. The model predicted less
rotation at the an er rotation at the
subtalar joint han th in the Th on of the
helical axis differed by no more than 45.68º at any joint. The model predicted centroidal
translation magnitudes at the ankle joint complex and ankle joint that were very close to
the experimental ured values difference f ankle co and 28.7%
difference for the ankle joint). The calcaneus translate ally relat the tibia in
m ent). The numerical pre
the experimental measurement of rotation at the ankle joint complex. The experimental
results showed that under the inversion moment, the hindfoot rotated primarily at the
th
kle joint than the experiment (50.6%) and much great
(-198.7%) t at measured experiment. e orientati
ly meas (-8.4% or the mplex
d medi ive to
122
both the model (22.69 mm) and experiment (17.90 mm) under the inversion load. The
model predicted calcaneus translations that were much more at the subtalar joint than the
experimental measurements (-754.5%). When comparing the pictures of the model and
the experiment, the modeled hindfoot appears to be inverted much more across the ankle
joint complex and particularly at the subtalar joint than its experimental counterpart
(Figure 38).
translations (Trans Mag, x, y, z) from neutral to inversion with the ATFL and CFL sectioned for the in vitro model and the test specimen
Ankle Joint Complex Ankle Joint Subtalar Joint
Table 25: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal
Parameter
Model Expt Diff Model Expt Diff Model Expt Diff
Angle ° 36.31 25.46 -
10.85 15.57 31.53 15.96 22.65 7.55 -15.10Rot Vec x 0.995 0.627 0.991 0.743 0.915 0.924
y -0.031 -
0.288 0.105 -
0.234 -0.044 0.160
z -0.091 -
0.724 45.68º
-0.079-
0.62740.42º
0.400 -
0.34845.64º
Trans
Mag mm 24.05 22.18 -1.87 2.78 3.90 1.12 21.79 2.55 -19.24X mm -7.91 12.79 20.70 -1.38 -0.40 0.98 -6.50 -0.07 6.43 Y mm -0.90 2.81 3.71 1.30 3.62 2.32 -2.30 -0.78 1.52 Z mm 22.69 17.90 -4.79 2.03 1.40 -0.63 20.67 2.42 -18.25
123
Figure 38:Comparison of inverted in vitro model (L) and corresponding MR image data (R) with the ATFL and CFL sectioned
In terms of the Grood and Suntay parameters, the calcaneus inverted relative to the tibia
( β AJC = -41.41º). The talus inverted at the ankle joint and the calcaneus inverted at the
subtalar joint by nearly equal amounts. (γ AJ = -18.67º,α STJ = 21.57º), (Table 26).
Table 26: Grood and Suntay parameters describing changes in In vitro hindfoot model kinematics from neutral to inversion with the ATFL and CFL sectioned
Parameters Ankle Joint Complex Ankle Joint Subtalar
Joint α º 0 -4.32 21.57 β º -41.41 4.25 -10.64 γ º 0 -18.67 -0.92
q1 mm -0.06 1.72 2.07 q mm 0.58 4.64 -0.27 2
q3 mm -2.85 -0.95 1.48
124
With the ATFL and CFL sectioned, an anterior drawer force caused similar responses in
centroidal translation at each joint magnitude in the model and the experiment (20.7%
difference at the ankle joint complex; 0.27% difference at the ankle joint; 33.8%
difference the subtalar joint), (Table 27). In the model, the majority of translation at the
ankle joint complex occurred anteriorly (x = -8.10 mm). In the experiment, translation
occurred both anteriorly and medially (z = 7.83 mm). This translation pattern was
peated at the ankle joint. At the subtalar joint, the model predicted a larger translation
m
were small (2.84 mm or less) in each case. The helical axis orientation calculated in the
model and the experiment was sim ankle joint complex differing by 30.08º. The
model predicted slightly greater helical axis rotation at the ankle joint than what occurred
in the experiment (-21.2% difference). The rotation at the ankle joint occurred about axes
with different orientations, though (63.66 difference in orientation). In the model, the
rotation occurred about the inferiorly oriented tibial inertial vector (y = 0.930), while in
the model the rotation occurred about an axis that was oriented posteriorly (x = 0.963). At
the subtalar joint, the orientation of the helical axes calculated in the model and in the
experiment differed greatly (108.74º orientation difference). More rotation occurred at
the subtalar joint in the experiment than that predicted by the model (33.8% more rotation
in the experiment). The positions of the talus and calcaneus in the experiment are
qualitatively similar to those predicted by the model under load (Figure 39).
re
agnitude than the experimental measurement (-36.5% greater), although the translations
ilar at the
º
125
Table 27: Screw axis rotations (Angle), helical axis orientation (Rot Vec x, y, z) and centroidal translations (Trans Mag, x, y, z) from neutral to anterior drawer with the ATFL and CFL sectioned
for the in vitro model and the test subject
Ankle Joint Complex Ankle Joint Subtalar Joint Parameter
Model Expt Diff Model Expt Diff Model Expt Diff Angle ° 0 9.59 9.59 11.65 9.62 -2.03 11.81 17.85 6.04
Rot Vec x -0.873 -0.534 0.368 0.963 -0.028 0.874
y 0.218 0.555 0.930 0.096 -0.834 0.459
z 0.436 0.638 0.0002-
0.251
63.66º
-0.551 -0.156
108.74º 30.08º
-
Trans 8.20 10.34 2.14 7.47 7.45 -0.02 2.84 0.76 -2.08 Mag mm X mm -8.10 -5.62 2 .43 -1.00 -0.02 0.98 .48 -7.05 -5.62 1Y mm -1.30 3.74 5.04 -0.10 1.72 1.82 -1.10 0.46 1.56 Z mm 0.07 7.83 7.76 3 -2 1.82 2.48 7.8 5.35 .42 -0.60
Figure 39: Comparison of the in vitro model (L) and corresponding MR image data (R) for anterior drawer with the ATFL and CFL sectioned
126
In t
omplex both in the anterior direction (q = -7.75 mm) and inferiorly (q3 AJC = -8.28
mm), (Table 28). The anterior translation occurred primarily at the ankle joint (q3 AJ = -
7.02 mm). The model al predict stantia al rota the talus relative to
the tibia (
terms of the Grood and Suntay parameters, translation occurred at the ankle join
c 2 AJC
so ed sub l intern tion of
β AJC =11.53º e as those reported for anterior
drawer testing with only the ATFL sectioned.
Table 28: Grood and Suntay Parameters describing changes in in vitro hindfoot model kinematics
s Complex Joint
). The results were nearly the sam
from neutral to anterior drawer with the ATFL and CFL sectioned
Parameter Ankle Joint Ankle Joint Subtalar
α º 0 -1.49 -1.99 β º 0 11.53 4.69 γ º .28 -10.63 0 -2
q1 mm 3.07 0.29 -0.37 q2 mm -7.75 2.08 0.02 q3 mm 8.28 -7.02 -2.08
127
SENSITIVITY ANALYSES
Ligament Orientation
Inversion ( β ) decreased by 6.5% as the CFL calcaneal insertion moved anteriorly 5 mm
relative to its original location (Table 29). As the calcaneal insertion moved 5 mm
osteriorly, inversion increased by 0.56% and as the insertion translated 10 mm
po
Table 29: The effect of CFL calcaneal insertion locatio n ankle joint complex kinematics from neutral to inverted position
AJC Originanterio mm
sterio mm 0 mm
p
steriorly, inversion increased by 10.9%.
n o
l 5A r o
5 P r
1 Posterior
α º 0 0 0 0 β º -12.4 -12.47 -11.65 -13.83 γ º 0 0 0 0
q1 mm -3.55 -4.55 -2.9 -2.07 q2 mm -0.17 -0.20 -0.17 -0.12 q3 mm -0.47 -0.48 -0.53 -0.6
Number of Model Ligament Elements
The amount of eversion occurring at the ankle joint complex ( β in Table 30) increased
after an additional force component was used to represent the TSL and TCL structures
( β increased by 15.9%). Adding a fourth force component did not substantially increase
calcaneus eversion relative to the tibia. ( β decreased by less than 1%).
128
Table 30: The effect of using multiple force components to represent the TSL and TCL ligament structures on ankle joint complex (AJC) kinematics in eversion
T s otal number of TCL/TSL component
AJC 2 3 4 α º 0 0 0
º 20.68 23.97 23.77 βγ º 0 0 0
q1 mm 5.83 5.5 5.71 q2 mm -0.23 -0.39 -0.379 q3 mm 6.058 4.71 6.458
TCL/TSL structures caused the talus to
ore (
Adding a third component to represent the
internally rotate m β increased by 14.2%) and to evert less (γ decreased by 22.5%)
at the ankle joint (Table 31). Adding the fourth force component affected the motion at
the ankle joint less ( β and γ affected by less than 3.6%).
Table 31: The effect of using multiple force co
structures on ankle joimponents to represent the TSL and TCL ligament
nt (AJ) complex kinematics in eversion
Total number of TCL/TSL components AJ 2 3 4 α º 15.02 16.00 16.03 β º 6.35 7.25 7.51 γ º 11.19 8.67 8.39
q1 mm 0.29 -0.26 -0.19 q2 mm -0.96 -0.37 -0.32 q3 mm -3.94 -4.77 -4.83
129
Adding a third force component to describe the TCL/TSL structures caused the calcaneus
to evert more at the subtalar joint than when two structures represented these ligaments
(Table 32) (α increased by 42.8%). Adding a fourth force component did not affect the
amount of eversion (α increased by less than 1%).
Table 32: The effect of using multiple force components to represent the TSL and TCL ligament structures on subtalar joint (STJ) kinematics in eversion
Total number of TCL/TSL components
STJ 2 3 4 α º -13.47 -19.24 -19.37 β º -1.43 0.17 0.2 γ º -12.18 -13.53 -13.73
q1 mm 3.35 4.30 4.34 q2 mm -4.89 -5.57 -5.72 q3 mm 0.41 -0.89 -0.84
The talus plantarflexed at the ankle joint (α = 10.6º) when one force com
represented the deep portion of the PTTL as shown in Table 33. When a second
component was added to describe the PTTL, the talus dorsiflexed (
ponent
α =
ponent cause the talus to internally rotate less
-4.39º) at the
ankle joint. The third force component did not cause further changes in talar
plantarflexion. Adding a third com
( β decreased by 18.6%) and to invert less (γ decreased by 6.5%) than with two
components.
130
Table 33: The effect of using multiple force components to represent the deep PTTL ligament on ankle joint kinematics from neutral to inverted position with ATFL sectioned and CFL sectioned
Total number of PTTL components
AJ 1 2 3 α º 10.60 -4.39 -4.13 β º 4.56 4.67 3.38 γ º -18.91 -19.09 -17.84
q1 mm 1.78 1.76 1.93 q2 mm 4.76 4.93 4.44 q3 mm -1.22 -1.23 -0.83
Linear Ligament Representation
The inversion-eversion load-displacement graph of the ankle joint complex (Figure 40)
for the in vitro model with linear ligament stiffness properties had a region of high
flexibility between ± 0.5 N-m of torque. The loading and unloading path differed in this
region. At higher torques, the flexibility rapidly decreased but still had a non-zero value
at the maximum torque (± 3.4 N-m).
131
-15
-10
-5
0-3.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 3.5
Torque [N-m]
β
Figure 40: Inversion [ + ] / eversion [ - ] load-displacement curve for in vitro model using linear ligament stiffness characteristics
5
10
[deg
]
Contact Damping
As the in vitro model’s contact damping term decreased the oscillation amplitude of the
ankle joint complex’s load-displacement curve as the joint moved from the high
exibility to the low flexibility region increased (Figure 41). The contact damping term did
n
on ate flexi changed slig
diff ent da
fl
ot affect the joint range of motion or late flexibility. The eversion loading path during
the transiti between the early and l bility regions htly for the
er mping terms.
132
-35
-25
-5
5
15
-3 -2 -1 0 1 2 3
ue ]
β [d
eg]
Figure 41: T ct o nta pin e p 0. , 1 he l
c = 0.1
c = 1.0
-15
25
c = 0.05
Torq [N-m
g paramhe effe f the co ct dam ter (Dam ing (c) = 05 , 0.1 .0) on t oad-
displacement characteristics of the ankle joint complex in inversion [ + ] / eversion [ - ]
133
PREDICTION STUDIES
Hindfoot Kinematics
e Effects of Ligament Removal
Removing the ATFL force from the in vivo model caused very small changes in the
kinematics of the ankle joint complex and the subtalar joint under an inversion moment
(Table 34). Sectioning the ATFL caused t
Th
he talus to evert at the ankle joint (γ AJ = -1.67
r the cut ATFL condition). Removing both the ATFL and CFL forces from the
mulation caused the talus to invert at the ankle joint and the calcaneus to invert at the
ubtalar joint
fo
si
(γ AJ = 30.52º and α STJ = -17.79º) for the cut ATFL as nd cut CFL
T in le of th
joint (
condition). his condition also resulted ss external rotation e talus at the ankle
β = c t th d
talus elative the ia inc sed 7 m to th ndi
Com ed A n re ls ed lus sl riorly (q2 =
m me t t le h n wa ted
Table 34: The effect of isolated rupture of the ATFL (cATFL) and combined rupture of the ATFL cs in inversion as predicted using the in vivo model
P
0.68º) ompared o othe er con itions. The amount of dorsiflexion of the
r to tib rea by 40 .2% co pared the o er co tions.
bin TFL a d CFL moval a o caus the ta to tran ate infe 5.71
m) d an dially (q1 = -5.33 mm) a he ank joint w en the hi dfoot s inver .
and CFL (cATFL+ cCFL) on hindfoot kinemati arameters Ankle Joint Complex Ankle Joint Subtalar Joint
Intact cATFL + Intact cATFL + Intact cATFL + cATFL
cCFL
cATFL
cCFL
cATFL
cCFL
α º 0.00 0.00 0.00 -2.33 -2.34 -9.15 -8.40 -8.40 -17.79
β º 8.29 8.30 47.92 5.54 5.54 0.68 -1.68 -1.67 -8.47 γ º 0.00 0.00 0.00 0.40 -1.67 30.52 -6.20 -6.20 -3.92
q1 mm 3.12 3.13 -5.15 0.55 0.55 -5.33 -0.94 -0.94 0.27 q2 mm 0.70 0.70 -0.64 -1.50 -1.50 5.71 -2.34 -2.34 1.36 q3 mm 0.70 0.70 6.54 1.59 1.59 0.04 2.61 2.61 5.42
134
Eliminating the ATFL from the in vivo hindfoot model caused minimal changes at the
ankle joint complex when the hindfoot was loaded in anterior drawer (Table 35).
xcluding the ATFL from the model led to the most prominent changes at the ankle joint
hen the hindfoot was in anterior drawer. For example, anterior translation of the
alcaneus from the
e translat s to increas nt
(q3 ncreas 2 xc ing T the model caused the talus to
dorsi x less
E
w
c relative to the tibia (q2) did not increase after the ATFL was removed
model; how ver this caused anterior ion of the talu e at the ankle joi
i ed by 0.3%). E lud the A FL from
fle (α decreased by 81.4%) at the ankle joint ed e i nd
and t nterna otate ore (
compar to th ntact co ition
o i lly r m β increa 38
Table 35: The effect of isolated rupture of the ATFL and combined CFL
hindfoot kinematics in anterior drawer as predicted using the in vitro model
Parameters Ankle Joint Complex Ankle Joint Subtalar Joint
sed by 0.4%).
rupture of the ATFL and on
Intact cATFL cATFL
+ cCFL
Intact cATFLcATFL
+ cCFL
Intact cATFLcATFL
+ cCFL
α º 0.00 0.00 0.00 -6.62 -1.23 -1.20 -2.24 0.39 0.98
β º 0.00 0.00 0.00 -1.02 -4.09 -5.14 3.33 2.24 2.51 γ º 0.00 0.00 0.00 -0.99 0.86 1.45 0.10 3.60 4.62
q1 mm 2.51 2.36 1.73 1.18 0.73 0.77 -3.72 -2.56 -2.23 q2 mm -7.77 -7.66 -7.97 -0.14 1.46 1.78 -1.47 -1.79 -0.96 q3 mm -0.28 0.35 0.07 -4.09 -4.92 -5.52 0.72 0.01 0.16
Removing only the ATFL force from the in vitro model caused very small changes in
each joint’s kinematics when the hindfoot was inverted (Table 36). At the ankle joint the
talus initially inverted slightly with all ligaments included in the model (γ AJ = 1.13º), but
after the ATFL was removed the talus everted slightly (γ AJ = -0.53º). After removing
135
both the ATFL and the CFL and inverting the hindfoot., the model predicted large
ll three joints ( β AJC = -41.41º, γ AJ = -18.67º and changes primarily in inversion of a
α STJ = 21.57º). The model predicted that the inversion occurring at the ankle joint
complex was divided almost evenly between inversion at the ankle joint and at the
subtalar joint. The talus also dorsiflexed more (α AJ increased by 77.0%) after both
structures were removed from the model.
hindfoot kinematics in inversion as predicted using the in vitro model
arameters Ankle Joint Complex Ankle Joint Subtalar Joint
Table 36: The effect of isolated rupture of the ATFL and combined rupture of the ATFL and CFL on
P
Intact cATFL + Intact cATFL + Intact cATFL + cATFL
cCFL
cATFL
cCFL
cATFL
cCFL
α º 0 0 0 -2.12 -2.44 -4.32 13.66 11.8 21.57
β º - -12.2 -41.41 -0.51 0.33 4.25 -1.87 -1.6 -10.64 12.47 γ º 0 0 0 1.13 -0.53 -18.67 0.99 0.19 -0.92
q mm -3.55 -3.95 -0.06 -0.15 1.72 1.52 2.2 2.07 0.01 1q2 mm -0.17 -0.18 0.58 -2.18 -1.74 4.64 4.2 4.77 -0.27 q3 mm -0.47 -0.51 -2.85 -0.35 -0.83 -0.95 0.63 0.17 1.48
Removing the ATFL from the in vitro model caused increased anterior translation of the
calcaneus at the ankle joint complex (q2AJC increased by 1.24 mm) under anterior drawer
loading. The increased anterior translation occurred primarily at the ankle joint (q3AJ
increased by 5.49 mm) (Table 37). Removing the ATFL from the model caused the talus
to internally rotate at the ankle joint, (
β AJ 11.53º) whereas in the intact condition the
talus externally rotated ( β AJ = -1.19º). This also caused the talus to dorsiflex less (α STJ =
136
7.23º with all ligaments and α STJ = -1.49º with no ATFL). Eliminating both the ATFL
odel led to no further chang
-
nd the CFL from the in vitro es in hindfoot kinematics.
Table 37: The effect of isolated rup he an bin re of th ATFL and C
hindfoot kinematics in anterior drawer as predicted using the in vit
Pa rs J om n nt btalar int
ma
ture of t ATFL d com ed ruptu e ro model
FL on
ramete Ankle oint C plex A kle Joi Su Jo
Intact cATFL c
cCFL I cA
c
ccA
cATFL
cCFL
ATFL+ ntact TFL
ATFL+
CFL Intact TFL +
α º 0 0 0 - - - -1.99 7.23 1.49 1.49 3.81 -1.99
β º 0 0 0 - 1 11.53 6.33 4.69 4.69 1.19 1.53 γ º 0 0 0 1.7 - -2.28 3.03 -10.632.28 -10.63
q1 mm 0.42 3.07 3.07 -0.01 -0.29 0.29 -4.61 -0.37 0.37 q2 mm -6.58 -7.82 -7.75 -0.12 2.08 2.08 0.02 2.2 0.02 q mm 3 6.47 8.28 8.28 -1.53 -7.02 -7.02 -1.18 -2.08 -2.08
137
Ligament Elongation, Strain and Forces (In vivo and In vitro Models)
The Effects of Ligament Removal
The in vivo hindfoot model predicted that under an inversion moment, the deep
component of the PTTL and the CFL experienced the greatest elongation (1.06 mm, 1.17
mm respectively), strain (0.21, 0.11 respectively) and load (74.54 N, 76.31 N
respectively), (Table 38). At the subtalar joint, several components of the interosseous
ligament (ITCL1, ITCL 9) and cervical ligament (CL1, CL4) elongated, experienced
increased strain and increased loading. The ITCL1 is a medial component of the
interosseous ligament while ITCL 9 is deeper in the tarsal canal. Excluding the ATFL
from the model caused minimal elongation, strain and force changes in the collateral and
subtalar ligaments during inversion. After the ATFL and CFL were removed from the
model, the deep portion of the PTTL, and the superior component of the PTFL
experienced greater elongation (1.47 mm, 4.33 mm respectively), strain (0.29, 0.22
respectively) and force (114.52 N, 131.15 N respectively). The elongation, strain and
force experienced by several interosseous structures (ITCL3, ITCL7, ITCL 9) and
cervical structures (CL1, CL4) also increased greatly (CL1 strain increased 8% and CL4
strain increased 16%).
138
Table 38: The effect of serial ligament sectioning on ligament mechanics in inversion for the intact, ATFL removed (cATFL) and combined ATFL and CFL removed (cATFL+ cCFL) conditions as
predicted by the in vivo model
Ligament Elongation [mm] Strain [mm/mm] Force [ N ]
cc
+c
Intact cATFL cATFL
+ cCFL
Intact cATFLcATFL
+cCFL
Intact ATFLATFL
CFL
ATTL -0.1 -0.103 0.57 -0.03 -0.03 0.18 0 0 25.4TSL -1.95 -0 -0.-1.95 -11.18 -0.08 -0.08 -0.45 .01 011 0TCL -0.36 -0.36 -6.8 -0.02 -0.02 -0.33 0 0 0
PTTLsuperf -0.53 -0.534 -3.54 -0.05 -0.05 -0.36 0 0 0PTTLdeep 1.06 1.074 74.54 7 111.47 0.21 0.21 0.29 4.54 4.52
ATFL -3.45 -3.46 9.4 -0.31 -0.31 0.85 0 0 0CFL 7 761.17 1.17 30.29 0.12 0.12 3.01 6.31 .319 0
PTFLinf 1.38 1.366 1.26 0.11 0.10 0.10 0 0 0PTFLsuper 1311.09 1.077 4.33 0.06 0.06 0.22 0 0 .15
ITCL1 0.78 0.776 3 36 10.66 0.32 0.32 0.27 6.03 .028 9.24ITCL2 0.58 0.58 0.22 0.13 0.13 1.13 1.0.05 133 0ITCL3 1.3 1.293 19.8 19. 341.47 0.32 0.31 0.36 799 .87ITCL4 0.43 0.43 0.32 0.11 1 10.11 0.08 .01 .01 0ITCL5 0.51 0.51 0.96 0.13 2.44 2. 170.13 0.24 435 .12ITCL6 -0.16 -0.162 -0.35 -0.04 -0.04 -0.10 -0.02 -0.02 -0.03ITCL7 0.41 0.407 3. 250.99 0.09 0.09 0.22 3.14 144 .24ITCL8 0.46 0.462 6.93 6. 60.42 0.10 0.10 0.09 936 .66ITCL9 1.26 1.259 16.58 15. 201.37 0.16 0.16 0.17 646 .69ITCL10 0.44 0.44 0.61 0.08 2.95 2. 50.08 0.11 954 .85ITCL11 0.63 0.63 1.31 0.08 1 10.08 0.17 .28 .28 8.1ITCL12 0.18 0.25 0.34 0.03 0.0.04 0.06 0 0 068
CL1 1.27 1.598 1 11 62.21 0.16 0.20 0.28 1.15 .15 1.44CL2 0.06 0.15 0.52 0.01 0.02 0.06 0 0 2.12 .12 .06CL3 0.23 1.26 0.66 0.04 0.24 0.13 1.07 1.07 9.39CL4 0.94 0.94 2.14 0.12 1 1 920.12 0.28 0.79 0.79 .145
139
The in vivo hindfoot model predicted that under an anterior drawer force, the ATFL
elongated (1.87 mm), which corresponded to a strain of 0.37 and a load of 52.41 N on the
lateral side of the ankle (Table 39). The ATTL and the deep PTTL resisted the anterior
drawer load medially. The ATTL elongated 0.61 mm, which corresponded to a strain of
0.20 and a load of 84.55N. The deep PTTL also elongated. Several force components
comprising the model of the interosseous ligament experienced the most elongation strain
and loads (ITCL1, ITCL6, ITCL7). The force components comprising the cervical
ligament were also involved in resisting the anterior drawer force, primarily CL 1 and CL
4. After removing the ATFL from the model, the ATTL, and the deep PTTL force models
experienced increased elongation compared to the intact condition. The CFL also
elongated, which corresponded to a strain of 0.17 (54.5% increase from intact condition)
and a force of 53.62N. At the subtalar joint, the load in the CL3 component of the
cervical ligament increased by 155.9%. Removing both the ATFL and the CFL force
components from the model caused further elongation of the deep PTTL (increase in
length of 0.5 mm) and the ATTL (increase in length of 0.05 mm). The increase in ATTL
length caused a large increase in this structure’s force (51.21N increase from cut ATFL
condition). Several interosseous structures elongated much more after the CFL was
removed from the model, particularly ITCL1 (0.13 mm length increase, .05 strain
increase and 97.1% increase force increase), ITCL 6 (0.2 mm length increase, .05 strain
increase, 116.7% force increase). ITCL 1 is a medial component of the structures
representing the interosseous ligament, while ITCL 6 is located deeper in the tarsal canal.
140
Table 39: The effect of serial ligament sectioning on ligament mechanics in anterior drawer for the bined ATFL and Cintact, ATFL removed (cATFL) and com FL removed (cATFL+ cCFL) conditions
as predicted by the in vivo model
Ligament E S
train [mm/mlongation [mm] m] Force [ N ]
Intact cATFL cATFL
+ cCFL
Intact cATFLcATFL
+ cCFL
Intact cATFLcA L TF
+ cCFL
ATTL 0.61 0.62 0.67 0.19 0.20 0.21 84.55 92.38 143.59TSL 1.00 1.51 1.71 0.04 0.06 0.07 1.41 6 94.1 3.7TCL 1.90 2.30 2.38 0.09 0.11 0.12 6.76 16.94 18.3
PTTLsuperf 1.88 -0.73 2.15 0.19 -0.07 0.22 10.52 10.30 6.54PTTLdeep 1.47 1.53 2.03 0.29 0.31 0.41 32.79 47.48 49.53
ATFL 1.87 4.15 4 0 0.14 0.17 0.37 0.37 52.51 0.0 0.0CFL 0.89 1.11 1.70 0.09 0.11 0.17 18.45 53.62 0.00
PTFLinf 0.58 0.50 0.46 0.04 0.04 0.04 0.00 0.00 0.09PTFLsuper -0.09 -1.00 -0.98 0.00 -0.05 -0.05 0.00 0.00 0.00
ITCL1 0.78 0.81 0.94 0.32 0.34 0.39 28.56 33.64 66.30ITCL2 -0.04 0.21 0 0 0.362 -0.01 0.05 0.08 0.0 0.0 0.00ITCL3 0.04 0.01 - -0.02 0 0 00.08 0.01 0.00 0.0 0.0 0.0ITCL4 0.22 0.32 0.48 0.06 0.08 0.13 0.00 00.0 1.39ITCL5 0.64 0.42 0.52 0.16 0.10 0.13 4.59 2 81.1 2.3ITCL6 0.62 0.54 0.74 0.17 0.15 0.20 19.80 14.43 31.27ITCL7 0.86 0.48 0.61 0.19 0.10 0.13 16.99 4.48 27.6ITCL8 0.00 0.09 0.08 0.00 0.02 0.02 0.00 80.8 0.76ITCL9 0.44 0.83 0.47 0.06 0.10 0.06 2.85 6 87.1 2.9ITCL10 0.29 0.18 0.22 0.05 0.03 0.04 1.45 70.6 1.02ITCL11 0.96 0.36 0 5 0.53 0.12 0.05 0.07 4.2 0.0 0.87ITCL12 0.75 0.36 0.50 0.13 0.06 0.09 4.75 0.57 751.
CL1 1.65 1.15 1.14 0.21 0.15 0.14 25.18 9.81 11.62CL2 -0.59 -0.67 -0.36 -0.07 -0.08 -0.04 2.01 2.08 1.62CL3 0.92 1.27 1.23 0.18 0.24 0.24 16.48 142.18 57.9CL4 1.66 1.68 1.72 0.22 0.22 0.22 40.83 42.48 45.39
141
The in vitro hindfoot model predicted that under an inversion moment, the CFL and the
ATTL experienced the greatest elongation (0.89 mm, 2.39 mm respectively), strain (0.22,
0.16 respectively) and load (58.48 N, 76.31 N respectively), (Table 40). At the subtalar
joint, the components representing the cervical ligament (CL1, CL2, CL3, CL4)
elongated, experienced increased strain and increased loading. No interosseous ligament
structures elongated when all ligaments were included in the in vitro model in inversion.
Removing the ATFL from the model caused negligible changes in the ligament
elongation, strain and force. With the ATFL and CFL structures removed from the model
the ATTL, deep PTTL and inferior PTFL elongated (0.12 mm, 0.89 mm, 2.17 mm
respective increases from cut ATFL condition), experienced increased strain (0.05, 0.16,
0.11 respective increases from cut ATFL condition) and increased load (46.06 N, 84.03
N, 139.01N respective increases from cut ATFL condition). Several components the
interosseous ligament also elongated (ITCL2, ITCL4, ITCL10, ITCL11). The structures
representing the cervical ligament (CL1, CL2, CL4) also elongated after both the ATFL
and CFL ligaments were removed from the model.
142
Ta y
Strai m]
ble 40: The effect of serial ligament sectioning on ligament mechanics in inversion as predicted bthe in vitro model
Ligament Elongation [mm] n [mm/m Force [ N ]
Intact cATFL cATFL
+ cCFL
Intact cATFLcATFL
+ cCFL
Intact cATFLcATFL
+ cCFL
ATTL 0.89 0.96 1.08 0.22 0.24 0.27 58.48 48.98 95.04 TSL -4.33 -4.14 -6.54 -0.19 -0.18 -0.29 0 0 0 TCL -3.89 -3.59 -8.84 -0.28 -0.26 -0.64 0 0 0
PTTLsuperf -1.02 -1.27 2.02 -0.08 -0.10 0.16 0 0 0 PTTLdeep 0.11 0.14 1.03 0.02 0.03 0.19 0.58 0.03 84.06
ATFL 0.36 0.44 7.3 0.08 0.10 1.60 11.95 0 0 CFL 2.39 2.20 24.57 0.16 0.15 1.68 1 1 07.39 04.95 0
PTFLinf 0.54 0.49 2.66 0.03 0.03 0.14 0.21 0.21 147.2 PTFLsuper 0.42 0.32 0.61 0.03 0.02 0.04 0.12 0.07 0.49
ITCL1 -3.39 -3.29 0.66 -0.77 -0.74 0.15 0 0 7.92 ITCL2 -3.15 -3.12 1.35 -0.48 -0.48 0.21 0 0 13.33 ITCL3 -2.59 -2.41 0.25 -0.36 -0.33 0.03 0 0 0 ITCL4 -2.17 -2.03 1.74 -0.30 -0.28 0.24 0 0 23.94 ITCL5 -1.98 -1.24 -1.34 -0.25 -0.16 -0.17 0 0 0 ITCL6 -1.52 -1.64 -0.43 -0.11 -0.12 -0.03 0 0 0 ITCL7 -0.97 -0.29 -0.51 -0.10 -0.03 -0.06 0 0 0 ITCL8 -1.99 -2.19 0.67 -0.18 -0.19 0.06 0 0 1.49 ITCL9 -1.86 -1.38 -1.84 -0.16 -0.12 -0.16 0 0 0 ITCL10 -1.53 -1.26 1.76 -0.17 -0.14 0.20 0 0 29.76 ITCL11 -1.08 -1.23 1.7 -0.09 -0.10 0.14 0 0 17.01
CL1 1.39 1.31 4.28 0.11 0.11 0.35 3.48 2.82 78.54 CL2 1.56 1.75 3.26 0.16 0.18 0.34 6.3 8.82 70.35 CL3 1.63 1.72 3.54 0.15 0.16 0.33 0.52 0.84 2.1 CL4 1.55 1 4.4 .45 0.12 0.11 0.34 2.06 1.64 45.15
143
The in vitro hindfoot model predicted that under an anterior drawer force, the ATFL
ad in the CL3 component of the cervical ligament increased by 155.9%.
emoving both the ATFL and the CFL force components from the model caused further
did not change ligament mechanics in the in vitro model.
elongated (0.91 mm), which corresponded to a strain of 0.20 and a load of 86.68 N on the
lateral side of the ankle (Table 41). The ATTL and the TSL resisted the anterior drawer
load on the medial aspect of the ankle complex. The ATTL elongated 1.24 mm, which
corresponded to a strain of 0.31 and a load of 73.52N. The TSL elongated 3.24 mm,
which corresponded to a strain of 0.14 and a load of 92.93 N. Several force components
comprising the model of the interosseous ligament experienced the most elongation,
strain and loads (ITCL1, ITCL). Only one force component comprising the cervical
ligament resisted the anterior drawer force, CL 1. After removing the ATFL from the
simulation, the ATTL, and the deep PTTL force models experienced increased elongation
compared to the intact condition. The CFL also elongated, which corresponded to a strain
of 0.17 (54.5% increase from intact condition) and a force of 53.62N. At the subtalar
joint, the lo
R
elongation of the TSL (increase in length of 0.21 mm, .01 strain increase), the ATTL
(increase in length of 0.05 mm, 0.02 strain increase) and the deep PTTL (increase in
length of 0.55 mm, 0.17 strain increase). The structure modeling the TSL experienced the
greatest load increase (38.8% load increase) following ATFL removal. Removing the
ATFL did not drastically increase elongation of the structures spanning the subtalar joint.
ITCL 6’s length increased by 1.52 mm, which corresponded to 0.11% strain and a
predicted ligament force of 9.99N. CL3 elongated by 0.99mm, which corresponded to
0.15 strain and a predicted load of 5.75N. Excluding both the ATFL and CFL structures
144
Table 41: The effect of serial ligament sectioning on ligament mechanics in anterior drawer as
ent Elongation [mm]
predicted by the in vitro model
Ligam Strain Force [ N ]
Intact cATFL cATFL
+ cCFL
Intact cATFLcATFL
+ cCFL
Intact cATFLcATFL
+ cCFL
A 9 TTL 1.24 1.33 1.33 0.31 0.33 0.33 73.52 44.89 44.8TSL 3.24 3.45 3.454 0.14 0.15 0.15 92.93 128.97 128.97 TCL 0.99 0.63 0 0.05 7.82 2.08 2.08 .63 0.07 0.05
PTTLsuperf 0.06 -0.52 -0.52 0.00 -0.04 -0.04 0.00 0.60 0.00 PTTLdeep -0.4 0.55 0.55 -0.07 0.10 0.10 -0.05 12.60 12.60
ATFL 0.91 2.03 2.03 6 . .00 9.22 9.22 0.20 8 .68 0 00 0CFL -4.81 -4.59 -4.59 -0.33 -0. 0.00 0.031 -0.31 0 0.00
PTFLinf 0.68 3 0.03 0.16 0.16 0.20 3.05 3.05 3.05 .05 PTFLs .37 -0.33 0.02 -0.02 -0.02 0.15 0.00 0.00 uper 0 -0.33
ITCL1 0.72 0.51 0 0. . 0 3. .43 0.51 .16 12 0 12 3 .69 1 43 13ITCL2 -0.86 -0 -0.13 -0.1 -0. 0.0 -0.01 -0.01 -0.85 .85 3 13 0 ITCL3 -0.25 -0.85 -0.85 -0.03 -0.12 -0.1 -0. 0.02 01 0 0.00 ITC .18 -1.66 -0.16 -0.23 -0.23 0.00 0.00 0.00 L4 -1 -1.66 ITCL5 -0.68 -0.09 -0.06 0 . .0 .00 -0.5 -0.5 - .06 0 00 0 0 0ITCL6 2.2 1.5 0.16 0.11 0.11 24.09 9.99 9.99 1.52 2 ITCL7 -1.25 -1.03 -1.03 -0.14 -0.11 -0.11 -0.02 -0.02 -0.02 ITCL8 1.05 -0.08 -0.08 0.09 -0.01 -0.01 8.93 0.00 0.00 ITCL9 0.64 0.76 0.76 0.05 0.06 0.06 4.16 4.80 4.80 ITCL10 -0.08 -1.26 -1.26 -0.01 -0.14 -0.14 0.00 0.00 0.00 ITCL11 0.06 -1.23 -1.23 0.00 -0.10 -0.10 0.00 0.00 0.00
CL1 0.6 -0.15 -0.15 0.05 -0.01 -0.01 0.73 0.00 0.00 CL2 0.69 1.68 1.68 0.07 0.17 0.17 0.32 5.75 5.75 CL3 1.24 1.57 1.57 0.12 0.15 0.15 -0.01 0.44 0.44 CL4 0.65 0 0 0.05 0.00 0.00 0.08 0.00 0.00
145
Joint Contact Force Magnitudes
Inverting the intact hindfoot caused the largest contact forces at the subtalar joint for both
the in vitro and in vivo models (Table 42). Inversion loading also resulted in loading at
e ankle for the in vitro model and to a lesser extent in the in vivo model. Sectioning the
TFL led to small changes in the contact forces at each joint for both models. Most
notably model.
Removing the ligament force representations for the ATFL and the CFL caused the
l subtalar odels to increase ( o subtalar joint load
increased by 150% and i ro subtalar jo ding increa 2719%). T vivo
model predicted that ankle joint loading would increase to 137.61 N from 1.63 N after
re ligament force representations from the model, however the in v odel
predicted that the ankle joint experienced ding with igaments re d in
inversion.
Loading the hindfo erior dr used force velop at t alar
joint (155.85 N in vivo in vitro) and to a lesser extent at the ankle joint (37.91
in vivo, 7.92 N in vitro), (Table 42). Contact between the talus and the fibula also
occurred in the in vivo model during the anterior drawer simulation (25.17 N in vivo).
Removing the ATFL from both models caused the ankle joint contact force to increase
(60.61N increase in vivo, 62.82 N increase in vitro). This did not greatly increase the
loads at the subtalar joint. Excluding both the ATFL and the CFL from the hindfoot
models caused the in vivo ankle and subtalar joint contact forces to increase (39.09N
th
A
, the loading at the ankle joint increased by 13.1% in the in vitro
oading in the joint of both m in viv
n in vit int loa sed by he in
moving both itro m
no loa both l move
ot mo n antdels i awer ca s to de he subt
, 100.17 N
N
146
ankle joint load increase, 50.61 N subtalar joint load increase). However, these forces did
not change in the in vitro model.
inversion and anterior drawer in the intact, ATFL sectioned (cATFL), and ATFL+CFL sectioned (cATFL+cCFL) conditions
Inv
Table 42: Contact force magnitudes at each hindfoot bone articulation for steady state loading in
ersion Anterior Drawer
Joint Intact cATFLcATFL
+ cCFL
Intact cATFL cATFL
+ cCFL
Ti-Ta 1.63 1.63 137.61 37.91 98.52 137.61
Ta-Fi 0.50 0.49 0.00 0.01 0.00 0.00 Ta-Ca 134.25 134.25 214.79 155.85 164.18 214.79
In Vivo [N]
Ca-Fi 0.00 0.00 0.00 25.17 4.36 0.00
Ti-Ta 52.6 59.50 0 7.92 70.74 70.74
Ta-Fi 0 0.00 0 0 0 0.00 Ta-Ca 110.52 11.44 322.5 100.17 106.47 106.47
In Vitro [N]
Ca-Fi 0 0.00 0 0 0 0.00
Ankle Joint Complex Flexibility Characteristics
The in vivo model predicted that the early flexibility was much greater than the late
flexibility in all motions (Table 43). When all ligaments were included in the model
flexibility and total flexibility. The early flexibility increased after the ATFL was
removed from the in vivo model in inversion (2.5% increase), eversion (5.6% increase)
and anterior drawer (4.2% increase). This did not affect late flexibility in any motion.
Removing both the ATFL and CFL ligament forces caused a drastic increase in inversion
(intact condition), the plantarflexion and dorsiflexion movements had the highest early
147
early flexibility (285.1% increase), total flexibility (276.4% increase) as well as large
increases in early flexibility for all other rotational motions. Late flexibility was not
affected by removing these structures from the model.
Table 43: Flexibility characteristics as predicted by the in vivo model for all motions
Flexibility Motion Condition Early Late Total
Dorsi [deg/N] Intact 17.48 0.73 9.11 Plantar [deg/N] Intact 18.23 0.31 9.58
Intact 7.87 0.58 4.23 cATFL 8.07 0.58 4.33 Inv [deg/N]
cATFL+cCFL 31.08 1.51 16.30 Intact 10.97 0.91 5.94
cATFL 11.58 0.77 6.18 Ev [deg/N] cATFL+cCFL 28.29 0.78 14.54
Intact 13.52 0.49 7.01 cATFL 13.52 0.48 7.00 Int [deg/N]
cATFL+cCFL 19.64 0.64 10.14 Intact 13.99 0.64 7.32
cATFL 13.98 0.64 7.31 Ext [deg/N] cATFL+cCFL 20.41 0.83 10.62
Intact 0.095 0.009 0.052 cATFL 0.099 0.008 0.054 q2 [mm/N]
cATFL+cCFL 0.095 0.010 0.053
148
The load-displacement graphs for each movement follow different loading and unloading
paths, and have high flexibility at low torque followed by a rapid non-linear decrease in
flexibility (Figure 42-45). Removing the CFL resulted in a drastic increase in ankle joint
complex range of motion in inversion (Figure 44) and internal and external rotation
(Figure 44). The inversion load-displacement plot (Figure 43) oscillated after the CFL
was removed in the low flexibility region.
-60
-40
-30
Torque [N-m]
condition
-50
-20
-10
10
20
30
40
-7.5 -5 5 7.5
α [d
eg]
Figure 42: Plantarflexion [ + ] / Dorsiflexion [ - ] load-displacement curves for in vivo model in intact
0-2.5 0 2.5
149
30
40
50CATFL
+ cCFL
0
-20
-10
10
20
- .5 -2.5 -1.5 -0 0.5 2.5 3.5
e [N-m]
β [d
eg]
Figure 43: Inversion [ + ] / eversion [ - ] load-displacement curves for model in int FL s and ATFL + tioned condi
3 .5 1.5
Torqu
in vivo
tions act, AT
ectioned CFL sec
-25
-20
-15
-10
-5 00
5
1
2
2
-3 1 2 3
Torque [N-m]
γ [d
eg]
Figure 44: Internal rotation [ - ] / external rotation [ + ] load-displacement curves for in vivo model in
intact, ATFL sectioned and ATFL + CFL sectioned conditions
cATFL
Intact
0
15
0
5
-2 -1
CA+
cCFL
TFL
cATFL Intact
150
0
1
2
3
4
5
6
7
8
0 25 50 75 100 125 150
Force [N]
q2 [m
m]
Figure 45: Anterior Drawer [ + ] load-displacement curves for in vivo model in the intact, ATFL sectioned and ATFL + CFL sectioned conditions
The in vitro model predicted that the early flexibility was much greater than the late
flexibility in all motions (Table 44). When all ligaments were included in the model
(intact condition), the movement flexibilities from highest to lowest were: internal
rotation (22.03 °/N-m), plantarflexion (17.54 °/N-m), external rotation (16.35 °/N-m),
and then dorsiflexion (16.21 °/N-m). The early flexibility increased after the ATFL was
removed from the in vitro model in inversion (19.6% increase), internal rotation (13.0%
ATFL from the model also decreased late flexibility in inversion (7.1% decrease),
eversion (37.2% decrease) and internal rotation(49.6%), although the magnitudes of these
cATFL
CATFL +
cCFL
Intact
increase), eversion (5.9% increase) and anterior drawer (6.7% increase). Excluding the
151
values were quite small. Removing both the ATFL and CFL ligament forces caused a
drastic increase in inversion early flexibility (113.4% increase) and total flexibility
(121.0% increase). External rotation early flexibility (47.4% increase), late (203.8%
increase) and total flexibility (50.7% increase) also increased. In all motions, the changes
in late flexibility were small.
Table 44: Flexibility characteristics as predicted by the in vitro model in all motions
Flexibility Motion Condition Early Late Total
Dorsi [deg/N] Intact 16.21 0.69 8.45 Plantar [deg/N] Intact 17.54 1.01 9.28
Intact 12.20 0.70 6.45 cATFL 14.59 0.65 7.62 Inv [deg/N]
cATFL+cCFL 31.13 2.56 16.84 Intact 13.50 1.21 7.36
cATFL 14.30 0.76 7.53 Ev [deg/N] cATFL+cCFL 23.99 0.75 12.37
Intact 22.03 1.21 11.62 cATFL 24.90 0.61 12.76 Int [deg/N]
cATFL+cCFL 40.73 3.22 21.98 Intact 16.35 0.52 8.44
cATFL 23.90 0.52 12.21 Ext [deg/N] cATFL+cCFL 35.22 1.58 18.40
Intact 0.089 0.008 0.049 cATFL 0.095 0.008 0.052 q2 [mm/N]
cATFL+cCFL 0.103 0.008 0.055
Each in vitro load-displacement plot (Figure 46 – 49) shows that the ankle joint complex
follows different loading and unloading paths, and has high flexibility at low torque
followed by a rapid non-linear decrease in flexibility. The flexion load-displacement plot
(Figure 46) sho low flexibility. ws oscillations in the transitions from high flexibility to
152
Removing the CFL resulted in a drastic increase in ankle joint complex range of motion
in inversion (Figure 47) and internal and external rotation (
Figure 48). The rotation load-displacement plot (
Figure 48) oscillated after the CFL was removed in the transition from high to low
flexibility. Removing the ATFL and CFL caused large increases in all rotations(Figure
47,
Figure 48) while removing the ATFL only caused an increase in the anterior translation
(Figure 49).
-40
-3
-20
-10-7.5 -5 -2.5 0 2.5 5α
condition
0
0
10
30
50
7.5
Torque [N-m]
[deg
]
Figure 46: Plantarflexion [ + ] / Dorsiflexion [ - ] load-displacement curves for in vitro model in intact
40
20
153
50
-20
-10
Torque [N-m
Figure 47: Inversion [ + ] / eversion [ - ] load-displacement curves for in vitro model in intact, ATFL
0-3.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 3.5
]
β [d
eg]
sectioned and ATFL + CFL sectioned conditions
10
20
30
40
-70
-60
-50
-40
-30
-20
0
20
-3.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 3.5
Tor -m]
[deg
]
rotation [ + ] load-displacement curves for in vitro model in intact, ATFL sectioned and ATFL + CFL sectioned conditions
cAT
L
FL
cATF+
cCFL
Intact
CATFL +
cCFL
10
30
-10
que [N
γ
Figure 48: Internal rotation [ - ] / external
cATFL
Intact
154
8
-1
0
1
2
0 25 50 75 100 125 150
Hindfoot Mechanics in Plantarflexion and Dorsiflexion
3
4
5
6
7q 2
[mm
]
Force [N]
Figure 49: Anterior Drawer [ + ] load-displacement curves for in vitro model in intact, ATFL
sectioned and ATFL + CFL sectioned conditions
Kinematics
The in vivo simulation predicted greater ankle joint complex range of motion in
dorsiflexion (α AJC= -50.82º) than plantarflexion (α AJC = 30.13º), (Table 45). These
motions occurred primarily at the ankle joint (α AJ= 19.80º plantarflexion,α AJ = -44.30º
dorsiflexion), and to a lesser extent at the subtalar joint ( β STJ = 8.19º plantarflexion,
β STJ = -5.93º dorsiflexion). Plantarflexion at the ankle joint complex was coupled with
external rotation of the talus relative to both the tibia. Dorsiflexion was also coupled with
inversion of the talus at the ankle joint and at the subtalar joint.
cATFL CATFL +
cCFL
Intact
155
Table 45: Kinematics of the in vivo hindfoot model in plantarflexion / dorsiflexion
P Ankle Jo Ankle Joint Subtalar Joint arameters int Complex Plantar Dorsi Plantar Dorsi Dorsi Plantar
α º 30.13 -50.82 19.80 -44.30 2.49 11.02
β º 0.00 0.00 5.03 0.36 8.19 -5.93 γ º 0.00 0.00 5.32 8.63 -7.40 1.52
q -1.36 -2.11 -2.55 3.32 1 mm 0.17 -1.54 q2 mm -0.26 3.46 -2 .00 2.44 -1.57 -0.34 q3 mm - - -1.51 0.25 1.00 2.60 2.51 0.55
Th simulation predic reater join plex plantarflexion (e in vitro ted g ankle t com α AJC=
42.67º) than dorsiflexion (α AJC = -35.53º), (Table 46). These motions occurred
prim the an int arily at kle jo (α AJ= 24.00º pl exionantarfl ,α AJ = -31.3 rsiflexio
to a lesser extent at the subtalar joint (
1º do n), and
β STJ = 16.87º plantarflexion, β STJ = -2.74º
dor ). The lso ext y rotated relative to both the tibia and the calcaneus
dur foot p flexion siflexio s co with l rotatio
inversion of the talus at the ankle joint and at the subtalar joint.
e 46: K tics of the in vitro hindfoot model i rflexio iflexion
P Ankle Joint Complex le J alar Joint
siflexion talus a ernall
ing hind lantar . Dor n wa upled interna n and
Tabl inema n planta n / dors
arameters Ank oint Subt Plantar Dorsi Plantar Dorsi Plantar Dorsi
α º 42.67 -35.53 -6.97 24.00 -31.31 5.91 β º 0.00 0.00 -2.74 -2.93 3.96 16.87 γ º 0.00 -40.00 1.26 -5.12 6.77 .92
q1 mm 0.00 0.00 0.94 -4.17 6.40 0.16 q2 mm 2.06 -0.04 -3.18 2.02 2.20 -1.41 q3 mm 4.41 0.04 4.05 -1.40 -0.18 0.08
156
Ligament Elongation, Strain and Forces (In vivo and In vitro Models)
The in vivo simulation of plantarflexion elongated, increased the strain and increased
loading of the ATTL (0.88 mm elongation, 0.28 strain, 450.79 N load), the deep PTTL
(1.90 mm elongation, 0.38 strain, 480.40 N load), and the ATFL (2.52 mm elongation,
0.23 strain, 112.78 N load), (Table 43Table 47). The CFL elongated 0.70 mm in the
platarflexion simulation. At the subtalar joint the ITCL1, ITCL 8 components of the
interosseous ligament model and the CL 3 component of the cervical ligament model
elongated.
The dorsiflexion simulation resulted in elongation of the ATTL (0.79 mm elongation,
0.25 strain, 251.18 N load) and the superficial component of the PTTL (3.35 mm
elongation, 0.34 strain, 507.04 N load). The ATFL was not loaded in dorsiflexion. At the
subtalar joint, ITCL1 and CL3 developed forces.
157
Table 47: Ligament mechanics in plantarflexion and dorsiflexion as predicted by the in vivo model
Lig P ament lantarflexion Dorsiflexion v IIn ivo n vivo
Elong
[m [mm/Fo Elongation
[Strain
[m ] e ation
m] Strain
mm]rce
[N] mm] m/mmForc[N]
ATTL 0.88 0.28 45 251.180.79 0.79 0.25 TSL 2.09 0.08 12.63 - - .014.01 0.16 -0TCL -1. -0 1.92 3921 .06 0.00 0.09 5.
PTTLsuperf -3.93 -0.40 3.35 .040.00 0.34 507PTTLdeep 1.90 0.38 4 1.03 7880.40 0.21 2.
ATFL 2.52 0.23 11 -0.39 - 002.78 0.04 0.CFL 0.70 0. 0 0607 6.79 .94 0.09 23.
PTFLinf 2.68 0.20 2.33 3317.30 0.18 5.PTFLsuper 3.46 0.18 3 -0.67 .002.34 -0.03 0
ITCL1 1.05 0.43 8 329.989.03 1.56 0.65 ITCL2 1.13 0.26 13.69 1.52 0.35 47.36ITCL3 1.24 0.30 14.29 1.41 0.34 25.41ITCL4 0.75 0.20 1.04 6.30 0.27 20.01ITCL5 0.33 0.08 .000.00 0.16 0.04 0ITCL6 0.56 0.15 1 -0.03 .025.68 -0.01 -0ITCL7 0.14 0.03 -0.57 - .000.06 0.13 0ITCL8 1.14 0.25 5 .829.07 0.42 0.09 5ITCL9 1.62 0.20 3 .311.08 1.40 0.18 21ITCL10 0.90 0.16 1 -0.24 .003.58 -0.04 0ITCL11 0.38 0.05 -1.91 .000.32 -0.25 0ITCL12 0.90 0.16 8.53 -1.37 -0.24 0.00
CL1 1.61 0. - 0020 25.84 2.42 -0.31 0.CL2 1.42 0. - - .0017 15.25 0.82 0.10 0CL3 1.91 0. 1 1 .4037 89.43 .43 0.28 60CL4 1.46 0. - .0119 28.58 0.83 -0.11 -0
158
The in vitro simulation of plantarflexion elongated, increased the strain and increased
loading of the ATTL (1.47 mm elongation, 0.37 strain, 292.83 N load), the deep PTTL
component of the PTTL (2.56 mm elongation,
0.20 strain, 136.38 N load) and the CFL PTTL (2.38 mm elongation, 0.16 strain, 367.65
(1.20 mm elongation, 0.22 strain, 390.53 N load), and the ATFL (1.38 mm elongation,
0.30 strain, 297.33 N load) (Table 48). The CFL elongated 1.20 mm in the plantarflexion
simulation. At the subtalar the CL 4 component of the cervical ligament model elongated
and experienced the highest load (203.11N).
The dorsiflexion simulation resulted in elongation of the ATTL (1.53 mm elongation,
0.38 strain, 397.18 N load), the superficial
N load). The ATFL elongated in dorsiflexion as well (1.29 mm elongation, 0.28 strain,
229.72 N load). At the subtalar joint, only ITCL 7 developed forces.
159
Table 48: Ligament mechanics in plantarflexion and dorsiflexion as predicted by the in vitro model
Ligament Plantarflexion Dorsiflexion
[mm] [mm/mmElongation Strain
]Force [N]
Elongation[mm]
Strain [mm/mm]
Force [N]
ATTL 1.47 0.37 292.83 1.53 0.38 397.18TSL -0.04 0.00 0.00 -0.79 -0.03 0.00 TCL -0.90 -0.06 0.00 0.06 0.00 0.22
PTTLsuperf -1.91 -0.15 0.00 2.56 0.20 136.38PTTLdeep 1.20 0.22 390.53 0.18 0.03 1.01
ATFL 1.38 0.30 297.33 1.29 0.28 229.72CFL 1.22 0.08 6.89 2.38 0.16 367.65
PTFLinf 1.72 0.09 3.30 2.63 0.13 27.41 PTFLsuper 2.88 0.19 299.80 1.87 0.12 16.36
ITCL1 -0.90 -0.20 0.00 -1.98 -0.45 0.00 ITCL2 -2.23 -0.34 0.00 -1.56 -0.24 0.00 ITCL3 0.09 0.01 0.65 -1.64 -0.23 -0.01 ITCL4 -1.60 -0.22 0.00 -1.07 -0.15 0.00 ITCL5 -0.39 -0.05 0.00 -0.17 -0.02 0.00 ITCL6 -1.46 -0.10 0.00 -3.23 -0.23 0.00 ITCL7 0.21 0.02 1.45 0.61 0.07 6.84 ITCL8 -1.25 -0.11 0.00 -3.50 -0.31 0.00 ITCL9 -0.55 -0.05 0.00 -1.29 -0.11 0.00 ITCL10 0.48 0.05 4.12 -2.22 -0.25 0.00 ITCL11 1.23 0.10 11.48 -2.60 -0.21 0.00
CL1 2.48 0.20 13.07 -1.41 -0.11 -0.01 CL2 4.42 0.46 203.11 1.01 0.11 1.88 CL3 4.49 0.42 42.16 0.10 0.01 0.00 CL4 2.72 0.21 18.52 -1.36 -0.10 0.00
160
CHAPTER 5. DISCUSSION
MODEL DEVELOPMENT
TM
The first portion of this project consisted of assembling and using tools (3DViewnix,
Marching Cube software, GEOMAGIC ) to develop patient-specific image-based
dynamic models of the hindfoot structure, including the bones and ligaments. This
method is applicable to the development of any joint model. The image processing,
geometric extraction and CAD development portion of this method is appropriate for
using in any mechanical analysis software, therefore these components could also be used
to develop finite element models of joints.
Future analyses of bone geometry changes due to bone surface smoothing and decimation
in Geomagic Studio should include analysis of local changes in bone geometry using the
principal axes data. Although bone volume changes were small, this term quantifies only
global changes in bone geometry. Fortunately, no articulating surfaces required local
surface smoothing and the articulating surface geometry was not drastically altered.
Therefore, it is likely that surface smoothing minimally affected joint mechanics.
MODEL ASSUMPTIONS AND LIMITATIONS
Boundary Conditions
Ankle Joint Complex Loading Constraints
Soft tissue deformation caused the in vivo experimental ankle joint complex loading
constraints to differ from the model’s. Soft tissue deformation caused the absolute motion
of the AFT to be greater than each bones’ actual motion. This is a fundamental limitation
161
of joint motion measurement using external arthometer devices. Applying consistent joint
fixation techniques for subject tests in different conditions allows the differences between
onditions to be determined accurately.
Although a rod was used to fix the subject’s heel to the foot plate of the ALD for in vitro
sMRI testing, the heel tended to rotate about the long axis of the rod. Therefore, rotations
occur in the in vitro experiment that the model does not predict. For example, in the
sMRI anterior drawer experiment, all rotations about all axes of the ALD are locked;
therefore rotations at the ankle joint complex should be zero. However, both the in vivo
and in vitro experimental results (Tables 12, 17) show that the calcaneus rotated relative
to the tibia in anterior drawer (2.36º in vivo, 8.72º in vitro). In the anterior drawer
simulations, all rotations about the ankle joint complex were locked in the neutral
position. Therefore, unlike the experiment, the in vitro and in vivo simulation predicted
no rotation at the ankle joint complex about the helical axis (Tables 12, 17) or about the
Grood and Suntay axes (Tables 13, 18). The disagreement between experimental and
numerical constraints also caused the orientation of the helical axis to differ. For
example, in the in vivo anterior drawer experiment, rotation occurred about a medially
oriented axis (Table 12, z = 0.892) indicating that the calcaneus dorsiflexed. This differs
simulation, where the orientation of the helical axis was
c
from the results predicted by the
not medial (Table 12, z = -0.150) and no rotation occurred at this joint.
162
Rigidly Constrained Fibula
The fibula and the tibia were rigidly constrained in six degrees-of-freedom for all models
in all simulations. The basis for this assumption was that the fibula undergoes small
translations (≤1.4mm lateral, ≤ 0.5 mm distal) and small axial rotations (≤ 3°) during
plantarflexion and dorsiflexion[113] when the tibia is fixed. This assumption eliminated
the need to use the material properties of the anterior and posterior tibiofibular ligaments
s well as the interosseous structure, whose mechanical properties have not been
ligament acts to primarily restrict
lantarflexion of the talus as it inserts from the calcaneus to the navicular under weight
Since all simulations are performed in the non-weight-bearing
condition, the spring ligament’s primary function is eliminated; therefore it may not play
a vital role in non-weight-bearing situations investigated in this study. The talo-navicular
joint is a highly mobile joint [55] and acts to transfer loads to the forefoot[116].
Therefore, in the non-weight bearing situations, studied in this model, it may minimally
restrain talus motion. Furthermore, during gait, the navicular constrains the talus when
the posterior tibialis tendon activates[114]. The model deals only with the non-weight-
a
documented. Furthermore, the tibia and fibula were mechanically grounded in all
cadaveric experiments; therefore locking the fibula in the model matches this boundary
condition.
Anterior Bone and Ligament Constraints
The model excludes the hindfoot’s distal structures including the bones (talus and
navicular) and soft tissues (spring ligament) therefore, the motion of the talus and
calcaneus may not be physiological. The spring
p
bearing[114, 115].
163
bearing, passive (no tendon forces) mechanics of the hindfoot; therefore the navicular’s
influence on hindfoot function may be minimal.
Bone (Inter-Cortical) Gaps
The bone surfaces were segmented at the level of the cortical bone; therefore the articular
cartilage was excluded from the model. This left inter-cortical gaps of up to 3.5 mm
between articulating surfaces. In order to have body-to-body contact the initial lengths of
the ligaments were shortened so that the ligaments generated a small force that closed the
inter-cortical gaps at the start of each simulation. The cartilage matches the shape of the
cortical bone layer; therefore excluding it will not alter the geometry of the articulating
rfaces. The increased space between bones may cause small increases in joint rotations
ontact Damping Coefficient
su
and translations. The talus is highly constrained on 3 sides by the tibia and fibula at the
ankle joint; therefore it is unlikely that the increased inter-cortical gaps will cause
dramatic changes in joint motion [32]. Unlike the ankle joint, the subtalar joint is
constrained by the ligaments[32], therefore, it is possible that the gaps will change
subtalar mechanics, particularly at the anterior talo-calcaneal articulation, where there is
only a small area of contact[67].
C
The contact damping coefficient was chosen to be small (1.0 N*s/mm in vivo, 0.1
N*s/mm in vitro) so that it did not dominate the dynamics of the model. Under this
experiment’s near quasi-static loading conditions (≤ 6 s loading times), the damping term
would not drastically effect model mechanics as shown in the contact damping
164
coefficient sensitivity study. The load-displacement curves obtained in this study (Figure
41) indicated that lower damping (0.05 N*s/mm) caused the oscillation amplitude to
increase slightly in the transition from the high flexibility to low flexibility region. The
entire range of damping values did not affect joint range of motion. In order to obtain an
improved estimate of contact damping, a simple viscoelastic model could be matched to
reviously published cartilage loading data. p
Several studies have explored the time dependant properties of cartilage[103, 105, 117],
but none reported these in terms of dynamic damping. The intervertebral disk is the only
biological material related to cartilage whose dynamic properties have been reported
[118]. The damping properties ranged from 0.032 N*s/mm at an axial loading frequency
of 20 Hz and 2.567 N*s/mm at a frequency of 5 Hz. In our study, each model’s contact
damping coefficient was within this range. Furthermore, each model’s damping term was
assigned the same value for all tests. Therefore, each model acts as its own control when
parameters are changed (i.e. ligaments are removed)
Contact Stiffness
The material properties of cartilage are typically presented in terms of Young’s
Modulus[105, 117]; a linear term that considers the geometry (cross-sectional area, initial
ngth) of the test specimen. ADAMS’ 3D contact force formulation requires a stiffness le
term, which must reflect the material properties of cartilage. In order to obtain stiffness
from the Young’s Modulus, it must be scaled by an area term and a thickness term
(Equation 17). The articulating surface contact areas vary when moving the
165
hindfoot[119]; therefore the stiffness term required by ADAMS’ would ideally vary as a
function of this parameter. Unfortunately, this feature was not included in the software.
cartilage’s Young’s Modulus in the model was the average area of
ceed a compressive axial strain of 100% (i.e. the
artilage cannot compress greater than its original thickness, which was between 2.43 and
modulus is higher than
that used in this study (0.374 MPa) under the more physiological loading conditions of
unconfined cyclic compression (Maximum Modulus = 65.7 MPa at 1 Hz). Therefore, we
underestimated the stiffness in our model because we based this value on the Young’s
The area used to scale
the polygons comprising each bone’s articulating surface (0.88 mm2) because ADAMS’
contact algorithm divides the geometry of each polygonal structure into the smallest
possible collection of polygons[96]. Therefore, the contact force will be dependent on a
small area of contact. This may underestimate contact stiffness, if an OBB encompasses
multiple polygons, but in this case the exponential term on penetration would still act to
provide sufficient resistant force to avoid large penetrations.
The non-linear compressive stress-strain properties of cartilage were based on the
assumption that cartilage cannot ex
c
3.5 mm for both specimens (Table 5) at the ankle and subtalar joints). Therefore, the
chosen penetration exponent (exponent = 9) caused the contact force to increase
drastically (Figure 21) as the penetration approached 2.6 mm of the original cartilage
thickness.
Recently published studies[105] indicated that cartilage’s elastic
166
modulus derived from equilibrium confined compression tests[104, 106]. These new data
also concluded that cartilage rarely exceeded 20% compressive strain even under the
most strenuous activities[105]; therefore, the exponent term used in this study was too
low. Future studies must incorporate this recently published data into the stiffness and
exponent terms for the contact description.
Ligament Mechanical Properties
The material properties of the collateral ligaments can deviate substantially from their
average values[39], therefore generalized load-displacement properties for the
gaments[102] may be inadequate for developing patient-specific predictions of joint
FL’s standard deviation in elastic modulus was ±65.1% of its
als
li
function. For example, the C
average value (512.0 ± 333.5 MPa)[39]. This indicates that there are substantial
variations in ligament mechanical properties across a population. Furthermore, the non-
linear load-strain characteristics used in this model were based on tensile testing of few
specimens (n=3)[102]. The ligament properties are also based on curve fits of elastic
response functions that are valid to no greater than 20% strain. The modeled ligaments
that experience strains greater than this will overestimate force, which would alter joint
mechanics. For example, overestimating ligament forces would cause late flexibility to be
less than what would occur biologically.
The ligament models did not include a relaxation expression because experimental data
shows that the ligaments will relax no more than 10% (at 10% step strain) over the time
periods under which loading occurred (≤ 3 s)[102]. If testing occurred over time interv
167
greater than this (≥ 10 s), a relaxation term should be included because the ligament
ated insertion
reas. The basis of this assumption was that the ATFL and both of these ligaments have
broad insertion areas and therefore may have similar physical structures[67].This may
ligament, the fibulotalocalcaneal ligament, and inferior extensor retinaculum. All of these
contribute some support to the subtalar joint in multiple motions and if included in the
model may significantly decrease subtalar joint motion[67, 120].
forces can decrease by greater than 25% [102].
Subtalar Ligaments
The mechanical properties of the interosseous talocalcaneal ligament and the cervical
ligament are undocumented; therefore their load-strain characteristics were estimated by
scaling the ATFL’s properties by the ITCL’s and the CL’s respective estim
a
cause them to have similar mechanical characteristics. The experimental comparison
indicated that the model over-estimated motion at the subtalar joint; therefore this
assumption may be inappropriate. To develop the model further, mechanical testing of
the subtalar ligaments is necessary.
The models may also have overestimated subtalar joint range of motion because several
ligaments, with documented anatomies[67], but with undocumented mechanical
characteristics were excluded from the model. These included the lateral talocalcaneal
168
Identification of Ligaments with Broad Attachment Areas
The technique used to identify the ligament insertions may not be appropriate for
structures that span larger areas, such as the TCL and TSL because the ligament force
vector directions may not represent those of the actual structures. The approach is
suitable for ligaments with linear structures such as the CFL because their orientation is
easy to visualize. When the TSL and TCL were represented by 3 elements, changes in
hindfoot kinematics were minimal in comparison to those occurring when both lateral
ligaments were removed. Therefore, it may be adequate to represent these types of
structures with 3 elements.
xperimental EquipmentE
ne major problem with in vivo flexibility measurement using the AFT is the fixation of
uring actual bone-to-bone motion.
O
the arthrometer across the hindfoot complex[60]. If the fixation is too tight, which is
necessary to minimize soft tissue movement, the patient will experience pain. If the
fixation is too loose, the measurements will be affected by soft tissue motion (slippage of
the heel fixation device or movement of the tibia). Therefore, true bone-to-bone motion
will not be measured and the apparent in vivo joint range of motion will be greater than
the internal motion of the bones. This may be one contributing factor to the overestimated
ankle joint complex range of motion when compared to that predicted by the model as
shown in Figures 31-33. The increased range of motion will also cause the total
flexibility measurements to decrease. Unlike the AFT, the stress MRI technique, using
the ALD, measures internal bone motion and not the movement of the arthrometer.
Therefore, stress MRI is more appropriate for meas
169
EVALUATION EXPERIMENTS The experimental testing tools (AFT, ALD) provided the means to obtain independent,
unique in vivo and in vitro mechanical data of the joint on which each model’s geometry
was based. No previously developed models of the foot and ankle provided this basis for
evaluating the results of their models. As summarized in Table 1, previous studies used
xial impulsive loading tests on one cadaver specimen[34-36, 72] or tested tarsal joint
sponse to axially loading the foot through the tibia[37, 38].
In vivo model
a
kinematics in re
The in vivo experimental load-displacement data did not exhibit high early flexibility and
a non-linear rapid decrease in late flexibility (Figure 31- 33) predicted by the model and
described previously[21, 60, 112]. The surrounding tissues (tendons, skin) of the patient,
which were excluded from the hindfoot model, likely decrease early flexibility in vivo.
Therefore, a model that excludes skin and tendon may be limited in predicting early joint
flexibility in vivo. The in vitro load-displacement results (skins, tendons and muscle
removed from specimen) more closely resemble the characteristics predicted by the
model (high early flexibility, non-linear transition to low late flexibility), (Figure 50).
170
-30
-20
-10
0
10
-12 -8 -4 0 4 8 12 16γ [d
eg]
20
30
Torque [N-m]
Figure 50: Sample in vitro flexibility test in internal rotation / external rotation
The support provided to the joints by the tendons and surrounding soft tissues, which are
excluded from the model, may decrease joint range of motion in vivo and explain, in part,
why the model tends to over-predict joint motions. For example, in inversion loading, the
predicted in vivo helical axis rotations were much greater at all joints than the
experimental rotations (Table 12). In anterior drawer loading, the predicted in vivo
centroidal translations were much greater at all joints than the experimental translations
(Table 14). However, the predicted in vitro ankle joint complex helical axis rotations in
the model for inversion were nearly the same as in the experiment (16.6% difference),
where much of the lateral tissues excluding the ligaments were removed (Table 17). The
redicted in vitro ankle joint complex centroidal translations were also nearly the same as
in the experiment (26.6% difference), (Table 19).
p
171
Overall, the motions measured in the in vivo experiment were small (Table 12, 14);
therefore calculating the percent difference et may be
misleadin the subtalar joint anterior translation (x-
direction, Table 14) differed from the experiment by 3.09 mm, but this corresponded to a
difference of 605.9%.
In vitro model
b ween the experiment and model
g. For example, in anterior drawer
The in vitr del captured the a comp xper l mo nt patterns.
For example, the predicted in vitro intact ankle joint complex helical axis rotations under
inversion loads wer the sa he expe
However, nk ontrib ajority of th tions he in vitro
inversion experime l condit n the c ry, th el p ed that the
subtalar joint contr the m f motion in inversion (Table 17, 21, 25). The
results indicate that the load-strain characteristi the ar li ts are too
flexible and further understanding of the ligamen chani opert a patient-
specific basis is necessary for accurate patient-specific predictions of joint function.
experiment and model agreed that ankle joint translation
o mo nkle joint lex’s e imenta veme
e nearly me as in t riment (16.6% difference), (Table 17).
the a le joint c uted to the m e rota for t
nt in al ions. O ontra e mod redict
ibuted to ajority o
cs of subtal gamen
t me cal pr ies on
The predicted in vitro ankle joint complex centroidal translations in anterior drawer were
also nearly the same as in the experiment (26.6% difference), (Table 19). Additionally,
the model over-estimated subtalar joint translation in the in vitro intact anterior drawer
simulation by 94.1%. There was good agreement between the intact in vitro model and
experiment for ankle joint translation in anterior drawer (3.8% difference), (Table 19).
Furthermore, both the in vitro
172
occurred primarily along the anteriorly oriented principle axis of the tibia (x) in the
experiment. After the ATFL was excluded
om the model, anterior centroidal translation magnitude increased by 196.4%, (from
direction of the applied load.
The formulation describing the ATFL’s mechanical properties may have been too stiff
compared to the actual structure because it played a much larger role in restraining ankle
joint anterior translation in the model than the
fr
2.52 mm to 7.47 mm), while in the experiment, it increased 48.8%. (from 2.62 mm to
3.90 mm), (Table 19, 23).
Inter-model Variability
The patient-specific models incorporated the morphological differences between subjects
(Table 49) and, as in the experiments; the hindfoot’s mechanical response to loading was
nsitive to them. Each test subject's morphological properties (bone geometry, ligament se
orientation, ligament initial length) vary (Table 49). For example, the principle axes
lengths of the patient’s talus were substantially greater (13.5 - 18.8% difference) than the
cadaver’s, while the lengths of the calcaneus were nearly the same (-2.5 – 3.1%
difference), (Table 49). In the experiment, specimens’s morphological uniqueness
contributes to their unique mechanical responses to loads. For example, at the ankle joint
complex the intact in vivo hindfoot model inverted 8.29° (Table 12) compared to 12.47°
for the intact in vitro model (Table 17).
173
Table 49: Comparison of the volume and principle axes lengths for the hindfoot bones of the in vivo and in vitro test subjects. Cal=Calcaneus, Tal=Talus, Tib=Tibia, Fib=Fibula, Diff=Difference
Principal Axis Lengths [mm]
Bone Specimen Volume [mm ] 1 2 3
3
in vivo 68928.7 85.7 39.2 29.1
in vitro 68842.2 87.9 39.9 28.2 Cal
% diff 0.1 -2.5 -1.7 3.1
in vivo 34702.9 58.1 39.3 22.4
in vitro 28634.7 51.2 33.1 19.2 Tal
% diff 21.2 13.5 18.8 16.7
in vivo 58808.3 58.2 42.1 35.3
in vitro 45778.0 55.6 32.8 32.0 Tib
% diff 28.5 4.6 28.2 10.3
in vivo 12962.9 58.2 25.7 17.1
in vitro 11853.7 62.8 19.0 15.5 Fib
% diff 9.4 -7.2 34.8 10.1
Similar to an experimental study, it is necessary to test a group of models, which reflects
biological variability within the human population and to perform statistical analyses on
the results. It is inappropriate to make generalized conclusions from a model without a
large enough sample size because the output data would lack sufficient statistical power.
Future work must include analyzing the output data for a sufficiently large group of
models (n = 10, for example).
174
Model Experimental Evaluation Overview
The comparison between experimental and numerical results are encouraging because the
model captured fundamental joint load-displacement characteristics (hysteresis in the
loading-unloading paths, high early flexibility, and a non-linear decrease to low late
flexibility) and predicted similar joint motion patterns with all ligaments included in the
model and with the ATFL and CFL removed.
The comparison between model results and experimental results also revealed each’s
adequacies. In vivo, the model predicted larger changes in inversion and eversion
e
ajority of the structural properties of the subtalar
ligaments were too flexible. The model predicted similar patterns of change in ankle joint
complex motion after the ligaments were removed; therefore the model is sensitive to
these changes in model parameters. However, the model over-estimated the effects in
ankle joint translation in anterior drawer after removing the ATFL. This indicates that the
load-sharing between the ligaments of the ankle joint may not have been physiological.
in
compared to the experiment. The experimental results may have been smaller than those
of the model because the model does not include the support provided by the surrounding
tissues of the hindfoot. These tissues may also have acted to decrease early flexibility,
while soft tissue motion of the AFT may have caused late flexibility to decrease.
In vitro, the model predicted changes across the ankle joint complex that close to those
measured in the experiment both for inv rsion and anterior drawer. Unlike the
experiment however, in the inversion simulation the subtalar joint contributed to the
m this motion. This indicates that
175
SENSITIVITY ANALYSES
Relatively large changes in the CFL’s calcaneal insertion (≥ 5mm anterior or posterior)
caused small changes in inversion range of motion (≤ 10.91%); therefore a small error in
locating its insertion during the ligament identification stage of model development
would have a small effect on the inversion range of m
otion. The CFL was tested in
version for this study because of its important role in stabilizing the hindfoot in in
inversion[53, 60]. The CFL reduced ankle joint complex inversion as its calcaneal
insertion moved 5 mm anteriorly ( β AJC decreased 6.57%) and increased inversion as it
moved 10 mm posteriorly ( β AJC increased 10.91%), (Table 29). These results also may
suggest that individuals who have a more vertically oriented CFL may be less susceptible
to inversion ankle injuries since they have reduced range of motion in this direction.
Those models that used anatomical atlases [35, 36, 38, 41, 42] to define the insertion of
the CFL may not be appropriate for predicting patient-specific hindfoot function, because
insertion location will affect model kinematics.
Hindfoot kinematics in eversion did not change drastically when representing the
TSL/TCL structure by more than 3 elements, therefore it is adequate to represent the
TCL/TSL structures by 3 force elements. The model was loaded in eversion for this
sensitivity analysis because the medial structures will be most active in restraining the
hindfoot in this direction[32]. Increasing the elements representing the TCL and TSL
from 2 to 3 elements increased eversion ( β AJC increased 15.91%), while increasing from
3 to 4 elements caused a negligible increase in eversion ( β AJC increased by 0.83%),
(Table 30). It may be necessary to represent the deep portion of the PTTL by at least 2
176
force components because large changes occurred in ankle joint kinematics in the ATFL
oved condition under an inversion moment (Table 31) when this structure’s
from 1 to 2 (
and CFL rem
components increased α decreased by 141.4%). The deep PTTL was tested
ditions (i.e. removing
gaments) are much larger than the changes introduced into the model by error in
under inversion in this condition because it may be important in stabilizing the hindfoot
after injury to the ATFL and CFL[19]. Less dramatic changes occurred at the ankle joint
after a third component was added. It may be most appropriate to represent this structure
with a force component at its proximal and distal ends because of its large insertion area
(0.79 cm2) and long insertion length (1.52 cm)[69].
Overall, the results of the sensitivity analysis suggest that the model can predict the effect
of rupturing the major passive support structures of the joint despite the error introduced
when identifying the ligament insertions or representing their geometry with linear
elements. Kinematic changes caused by altering model con
li
representing ligament geometry. For example, much smaller kinematic changes occurred
in the CFL inversion sensitivity analysis (∆ β ≤ 10.91%) than the change that occurred
when removing the CFL from the model (∆ β =477% change in vivo (Table 34),
∆ β =239.4% change in vitro (Table 36)).
Some fundamental load-displacement characteristics exhibited by all joints (hysteresis in
the loading-unloading paths, high early flexibility, and a non-linear decrease to low late
flexibility) are not solely caused by the non-linear load-displacement properties of
ligaments and cartilage. The inversion load-displacement characteristics of the in vitro
177
hindfoot model using linear ligament stiffnesses were qualitatively the same as the
nsitivity analyses indicate that some structures may be more appropriately modeled by
more than 1 structure such as the deep portion of the PTTL and the TSL/TCL. Changes in
hindfoot model using non-linear ligament load-strain elements (Figure 40). No previous
models have explored the relationship between ligament properties and joint load-
displacement properties. Although studies have documented these characteristics in
joints[29, 60], none investigated their causes.
No previously developed foot and ankle models explored the effects of various
assumptions made in developing their models (ligament insertion location, ligament
representation, material properties) through sensitivity analyses. The results of our
se
joint kinematics were less drastic after 2 components represented the deep PTTL and 3
components represented the TCL/TSL structures. Therefore, those models that used
multiple components to represent these structures [72, 76]may predict more stable
changes in hindfoot kinematics than those that did not.
178
MODEL PREDICTIONS
Kinematics
Inversion / Eversion
Like previous experiments, the models predicted that both the ankle joint and the subtalar
lly more of
e rotation in the models (73.4%)[53]. The comparison of predicted model kinematics
joint contributed to the entire motion across the ankle joint complex[53, 72, 76] in all
measured motions (inversion / eversion, plantarflexion / dorsiflexion, anterior drawer),
(Tables 34 - 37). The model predicted that the contribution of the ankle joint to inversion
/ eversion, was less than the subtalar joint (4.8% in vivo model, 9.1% in vitro model).
Unlike previous experiments[53], the subtalar joint contributed to substantia
th
with the literature indicates that further investigation of the subtalar joint’s ligament
material properties is necessary. This may also indicate that the motions characteristics of
the tested subjects were abnormal from the population averages.
The models over-estimated changes in ankle joint complex inversion after sectioning the
ATFL and CFL. For example, sectioning these ligaments caused inversion to increase by
13.5° on average [25], while the in vivo and the in vitro models predicted an increase of
39.63° and 28.94° respectively (Tables 34, 36).
179
Anterior Drawer
ivo models greatly overestimated by 176% to 300% the amount
f ankle joint dorsiflexion compared to an in vivo study of 18 subjects[112]; therefore,
t
eport higher levels of dorsiflexion (24.68 ± 3.25°)[53] than the
vivo experimental results reported above. These results from previous studies are still
substantially smaller than the results predicted by both models. They are also
substantially greater than the mean dorsiflexion (18.1± 6.9°) of a population of 300
After sectioning the ATFL, the models predicted changes in talar anterior translation at
the ankle joint that were similar to previous experiments[25]. In anterior drawer,
sectioning the ATFL caused the ankle joint anterior translation to increase an average of
3.1 mm[25], while the in vivo and the in vitro models predicted no increase and 5.49 mm
increase respectively (Tables 35, 37). The in vivo model may not have predicted an
increase in this motion because the anterior drawer force was directed laterally so that the
talus tended to wedge against the fibula, which resulted in small contact forces at this
articulation (Table 42).
Plantarflexion / Dorsiflexion
Both the in vitro and in v
o
the load-displacement properties of ligaments constraining the hindfoot in dorsiflexion
(PTTL) may be too flexible. Average dorsiflexion under 7.5 N-m orque was 12.7 ± 5.1°
in the experiment[112] while the in vivo and in vitro models predicted 50.82° and 35.53°
respectively. The surrounding soft tissues present in vivo (skin, tendon, muscle), which
may reduce joint range of motion and were excluded from our model also may contribute
to this large difference Although other in vitro studies do not mention the load applied to
produce the motion, they r
in
180
patients and would fall into the hypermobile classification (≥ 31.9°) in this study[121].
Since the PTTL is responsible for constraining the joint in dorsiflexion [32, 69], future
studies may be needed to better understand its mechanical properties.
The in vivo model predicted within one standard deviation the average in vivo
plantarflexion (31.0 ± 4.4°) of 31 subjects[112], while the in vitro model was within 3
standard deviations. In vitro experimental studies[53] measured greater hindfoot range of
motion in plantarflexion (40.92 ± 4.32°) than the in vivo experimental study mentioned
above. This value closely corresponded to that of the in vitro model (42.67°).
Like the results of previous experiments[53], the in vivo and in vitro models predicted
at the movements of dorsiflexion / plantarflexion occur both at the ankle joint and the th
subtalar joint, but primarily at the ankle joint. Furthermore, the models predicted that the
subtalar joint contributed more to total plantarflexion (Table 45,Table 46) than measured
in a previous experiment[53], indicating that the ligaments supporting the subtalar joint
were too flexible in the model. In the in vivo and in vitro models, ankle joint dorsiflexion
corresponded to 87.2% and 88.1% of total dorsiflexion respectively and 65.7% and
56.2% of total plantarflexion respectively. In the experiments, ankle joint motion
corresponded to 79.6% of total dorsiflexion and 80% of total plantarflexion.
181
Flexibility
Like previous experimental studies[53, 60, 112], the ankle joint complex of the in vitro
and in vivo models had highly non-linear load-displacement properties in all directions.
nder low loads around the hindfoot’s neutral position, the ankle joint complex was
external rotation, dorsiflexion,
version then inversion.
d
44).
.
U
highly flexible and at increasing loads the primary angular or linear displacements
reached asymptotic values (i.e. the joint’s maximum range of motion). Like previous
experiments, all motions had the highest flexibility in the neighborhood of the neutral
position (i.e. early flexibility)[60]. The hindfoot is most flexible in plantarflexion /
dorsiflexion, followed by internal rotation / external rotation and finally in inversion /
eversion[53]. The in vivo model followed this pattern, but the in vitro model was most
flexible in internal rotation followed by plantarflexion,
e
Unlike previous in vitro experimental studies[60], both models pre icted a drastic
decrease in late flexibility compared to early flexibility (Table 43, 44). Late flexibility
dropped by a factor of at least 10 in all motions in the models (Table 43, 44). In contrast,
in vitro experiments reported a drop by a factor of 3 at most[60]. For example, in the
anterior drawer experiments, early flexibility (0.06 mm/N) was nearly the same as late
flexibility (0.07 mm/N), unlike the model, which predicted a factor of 10 flexibility
decrease (Tables 43,
182
T ct
formulation used e the ligam character the
, soft n ay incre o
flexibility. Strain values greater than 20% will cause high ligament forces due to the
e te n the ad-st n fo lation[1 ] and rease te flex ty. Th
v vitr odel igam str ata in es th ver ructur the
and subtalar joints do exceed this value in inversion (PTTL deep, ITCL1 o; ATTL
i d rior d wer TL, TTL superf, PTTL deep, ITCL1, CL1 in vivo;
L vitro), (Table 38-41). When distance betw n liga inse n
points increases beyond 20% strain, the ligament tension increases drastically, which
mits range of motion and may decrease late flexibility.
ific knowledge of the hindfoot ligament mechanical properties may
r all ligaments and the external load. Since the
hindfoot is held together by greater than 6 ligaments, optimization criteria would have to
wo fa ors may influence the predicted low late flexibility: 1) the non-linear exponential
to describ ent load-strain istics and 2) in
experiments tissue motio m ase range of m tion and therefore increase late
xponential rm i lo rai rmu 02 dec la ibili e in
ivo and in o m ’s l ent ain d dicat at se al st es at ankle
in viv
n vitro) an a ent ra (AT P
ATTL, ATF in the ee ment rtio
li
Lack of patient-spec
contribute to the disagreement between the early flexibility measured in the models and
presented in previous experimental studies[60]. Inversion early flexibility was closest to
the experimental data. The in vivo model underestimated it by 22% while the in vitro
model overestimated it by the same amount. Both models over-estimated early flexibility
in all other motions by at least 50% compared to the experimental average (n=6) [60].
Modeling a larger database of subjects would help to indicate whether ligament structural
properties must be defined on a patient-specific basis. In order to determine ligament
structural properties on a patient-specific basis in vivo, it would be necessary to develop
the equations of static equilibrium fo
183
be d f
equations would have to be solved at mu long the tain
the ligaments’ non-linear structural properties.
Rem e or
the in vitro and in vivo models compared to previous experimental data[60] (ATFL vs.
Intact column in Table 50 below). For example, in anterior drawer, the in vivo and in
vitro model predicted 4% and 7% respective increases in early flexibility (Table 43, 44).
Previous experiments reported contradictory results; one reporting an increase of 109% in
early anterior drawer flexibility after ATFL sectioning [60]and another reporting that
changes only occurred in a 10° dorsiflexed position[29].
fter removing the CFL, the model predicted early and total flexibility changes that
l studies[60] (Table 50). The changes
more severe. For example, removing the ATFL
inclu ed to account for the additional structures. Furthermore, these systems o
ltiple increments a loading path to ob
oving the ATFL led to small or no increase in early flexibility in all movem nts f
A
followed the same pattern as previous experimenta
predicted by the model, however, were
and CFL structures from both models caused large changes in ankle joint complex early
flexibility and total flexibility in all motions compared to the intact condition (Table 50).
The largest changes were in inversion for both the model and experiment [60]. The
results again indicate disagreement between the modeled ligament properties and those of
the actual ligament.
184
Table 50: Percent difference in ankle joint complex flexibility between the three test conditions for
ATFL vs
Intact [%] Double cut vs
Intact [%] Double cut vs ATFL [%]
the in vivo and in vitro models and an in vitro experiment (exper)[60]
Test Condition in
vitroin
vitroin
vitroexper in vivo exper in vivo exper in vivo
early 109 4 7 141 0 16 15 -4 8 AD total 60 4 6 74 2 12 9 -2 6 early 13 3 20 86 295 155 64 285 113 Inv total 7 2 18 57 285 161 47 276 121 early 11 6 6 5 158 78 13 144 68 Ev total 6 4 2 15 145 68 8 135 64 early 12 0 13 8 45 85 19 45 64 Int total 2 0 10 6 45 89 4 45 72 early 14 0 46 24 158 115 9 144 47 Ext total 8 0 45 29 145 118 19 135 51
Ligament Loading
A comparison between the strain in several ligaments (ATFL, CFL, TCL) predicted by
the in vivo and in vitro models and a previous experiment[122] indicated that the
formulations used to model these tend to over-estimate strain output (Table 51). Both
models over-estimated strain levels in flexion, particularly at larger angles The predicted
strain patterns in flexion, however, were in agreement with experimental patterns[122].
Like the experiment, both models predicted increasing ATFL strain with plantarflexion
(Table 51). The models predicted increasing strain in the CFL with dorsiflexion (Table
51). The models also predicted small strains in the TCL, in dorsiflexion and laxity in the
TCL during plantarflexion (Table 51).
185
Table 51: Comparison of experimental[122] and predicted in vitro and in vivo percent strain as a
ATFL [% strain] CFL [% strain] TCL [% strain]
function of plantarflexion [-] and dorsiflexiuon [+] angle
Angle
°
Exp In vivo In vitro Exp In vivo In vitro Exp In vivo In vitro
30 -3.2 -9.49 16.2 1.9 5.60 11.4 1.2 3.80 2.71
20 -3.9 -10.21 4.46 .2 7.27 5.41 1.25 3.47 -2.98
10 -3.1 -11.15 -0.15 -.9 6.39 3.5 1.3 1.62 -4.31
0 -2.7 0.06 0.11 -1.9 0.32 0.95 1 -1.96 0.04
-10 -1.1 2.38 0.06 -2.9 -1.92 -2.65 -0.1 -5.66 -2.72
-20 0.5 7.46 2.28 -3 1.32 -0.03 -1.3 -8.01 -8.26
-30 1.9 20.78 13.37 -2.2 6.55 1.02 -2.6 -8.43 -10.22
PRELIMINARY CLINICAL SIGNIFICANCE Joint Load-Displacement Properties
The hindfoot models capture the fundamental load-displacement characteristics
(hysteresis in the loading-unloading paths, high early flexibility, and a non-linear
decrease to low late flexibility) exhibited by all joints[29, 60], and the results indicate that
hysteresis in the neutral region and low late flexibility are dependant on both contact of
the articulating surfaces and the strain in the ligaments. The preliminary discussion of one
model in the following paragraph suggests that the hindfoot’s load-displacement curve
exhibits hysteresis (different loading and unloading paths) in the high flexibility region
contact points when the joint is loaded because the articulating surfaces have different
186
and unloaded (Figure 51, Table 52). The hindfoot has low late flexibility because the
bone geometry constrains the joint from further movement. This causes the tibio-talar
contact force to rapidly increase in the late flexibility region (Table 52), but causes only
small changes in position (Figure 51). No other experimental or numerical studies
investigated the fundamental reasons for the load-displacement properties of joints.
Non-linear ligament mechanical properties are not the only cause of the joint load-
displacement characteristics described above. Naturally, the ligaments must also act to
constrain joint range of motion because when they are injured, joint kinematics change
(i.e. large increase in inversion range of motion after CFL injury)[32]. However, the in
vitro model’s inversion load-displacement characteristics with linear ligament stiffnesses
we s
igure 40).
T e
magnitude and contact location (Figure 51, Table 52) For example, as the in vitro model
everts, there are low contact forces in the neutral eversion loading zone (Point 1 in Table
52). A the e , t o o
51) the co e n f
conta shift 6.9 Table 52 Point 1 to Point 2b). e ion ent
incre s fro nts 2 (Figure 51 and Table 52) the tibia and fibula constrain the
talus seen e in in c for thes ul (4 tibi N
fibul and t o not it to furt
(2a, Tabl n pos 2). A ver oment decre om ts 3 the
re qualitatively the same as when the ligaments had non-linear mechanical propertie
(F
here is a correlation between hindfoot load-displacement properties, the contact forc
s version moment increases h oe hindf t everts rapidly (Point 1 t re2, Figu
ntact forc at the a kle joint increases rapidly (8.4 N) and the location o
ct s by mm ( As th evers mom
ase m poi to 3
as by th crease ontact ces at e artic ations 0.7 N a, 0.4
a), hey d allow evert her. Note that contact occurs in two locations
2b, e 52 i ition s the e sion m ases fr poin to 4,
187
contact force exerted by the talus on the tibia and fibula decrease (-40.6 N tibia, -0.4 N
ct force continues to
crease from points 6 to 7 as the inverting moment increases, with only small position
own in the plot of the tibio-talar
contact force magnitude in Figure 51. This may also cause oscillations in bone position at
the transition from high flexibility to low flexibility as seen in the in vitro load-
displacement graphs (Figure 46-48). These oscillations were related to the contact
damping term as shown by the contact damping sensitivity study (Figure 41). This non-
physiological artifact may be eliminated if the damping coefficient were closer to the
actual dynamic properties of cartilage.
fibula) but only small changes in tibio-talar contact location occur (0.2 mm) and the talus
loses contact with the fibula. From points 4 to 5 (Figure 51) the hindfoot moves back
towards the neutral position and the talus loses contact with the tibia (Table 52) as the
eversion moment decreases to 0 N-m. The hindfoot inverts rapidly in the neutral zone
(points 5 to 6, Figure 51) under low ankle joint contact forces (points 5 to 6, Table 52)
and with minimal inverting load. The talus re-contacts the tibia (10 N force increase,
Table 52) in the same location as position 5. The tibio-talar conta
in
changes and a slight shift (0.1 mm) in contact location. As the inversion load decreases
from points 7 to 8, the contact force also decreases (Figure 51) while the position of
contact remains unchanged (Table 52)
Rapid changes in the contact force magnitude (Figure 51) occured in the model as the
bone switched contact points in the neutral region. The initial contact between
articulating surfaces is characterized by the force spike sh
188
-15
-10
Torque [N-m]
Figure 51: In vitro model inversion / eversion load-displacement plot overlayed with tibio-talar (Ti-Ta) contact force magnitude. Points marked 1-8 on the load-displacement curve are specifically
Table 52: A
-5
0
5
10
15
20
-3 -2 -1 0 1 2 3
[deg
], Fo
rce
[N]
discussed.
nkle joint contact locations and force magnitudes [mag] for positions 1-8 of the load-
displacement curve shown in Figure 42
Articulation
β
Ti-Ta Ta-Fi
Contact Location [mm] Contact Location [mm] Point
X [mm] Y [mm] Z [mm]
Contact Force
Mag [N] X [mm] Y [mm] Z [mm]
Contact Force
Mag [N]
1 83.9 65.9 56.7 0.2 94.3 78.9 42.4 0.0 2a 77.5 68.3 58.1 0.1 2b 80.9 66.6 59.8
8.6
3 83.6 66.1 56.5 49.3 94.2 79.0 42.2 0.4 4 83.8 66.0 56.6 8.7 0.0 5 83.8 66.0 56.6 0.0 0.0 6 83.8 66.0 56.6 10.0 0.0 7 83.9 66.0 56.7 52.2 94.3 78.9 42.4 0.0 8 83.9 66.0 56.7 9.9 0.0
Ti-Ta Contact Force Magnitude
Eversion
4
5
6 1
Inversion
3
2
78
189
Ligament Loading
The model predictions indicate that ligament-loading patterns are sensitive to changes in
odel parameters (i.e. removing the ligaments). The load on the deep PTTL and the
ith the clinical observation that individuals with a history
f chronic instability also have associated injury to the PTTL [19].
m
cervical ligament greatly increased in inversion after removing the CFL and ATFL (39.98
N increase in vivo, (Table 38); 84.03 N increase in vitro, (Table 40)). This indicates that
these structures are more susceptible to injury in the presence of lateral ligament injuries.
This numerical finding agrees w
o
The sensitivity of ligament loads to changes in parameters is encouraging for future
studies of ankle joint fusion, subtalar joint fusion and ankle joint replacement. After
thorough model evaluation, future studies might focus on designing and evaluating ankle
replacements that closely reproduce ligament-loading patterns in order to avoid soft
tissue complications[6]. These hindfoot models might also be used to determine joint
fusion positions that minimize changes in ligament loading, in order to avoid subsequent
tissue degeneration due to abnormal ligament forces [9].
190
CHAPTER 6. SUMMARY AND CONCLUSIONS
OBJECTIVE
The main objective of this study
was to develop a subject specific (n=2: 1 in vitro, 1 in
ivo), three-dimensional dynamic model of the hindfoot using 3D sMRI data and evaluate
its ability to capture a wide range of mechanical phenomena including the mechanics of
h global and local bone smoothing operations and point dessimation
lgorithms were used to obtain bone surface representations (STL format) appropriate for
mechanical data of the joint on which each model’s geometry
v
the non-pathologic hindfoot and the mechanics of the hindfoot with ligament injury.
MODEL DEVELOPMENT
Existing software (3DViewnix) was incorporated with in-house software (Marching
Cubes program) to obtain the bone geometric information of the test subjects. These data
were then imported into a reverse engineering software package (GEOMAGIC Studio
5.0) in whic
a
modeling.
3DViewnix was also used to identify the ligament insertion points for the modeled
ligament structures. The ATFL, CFL, ATTL and TCL were represented with one linear
element. Other structures (PTFL, PTTL, TSL, ITCL, CL) were represented with multiple
elements because of their wider insertion areas. The forces generated were divided evenly
between all structures with multiple elements.
The experimental testing tools (AFT, ALD) provided the means to obtain independent,
unique in vivo and in vitro
191
was based. No previously developed models of the foot and ankle provided this basis for
Estimates for the structural properties of the collateral ligaments were obtained directly
from an existing experimental study[102]. The subtalar ligaments’ structural properties,
which are unknown, were estimated by scaling the load-strain properties of the ATFL by
the areas of the ITCL and CL.
The contact stiffness was estimated by scaling the documented compressive properties of
cartilage[104, 106] by the average area of the polygons representing the articulating
surfaces as well as the thickness of cartilage at the ankle joint and the subtalar joint. The
non-linear material properties of cartilage were modeled under the assumption that
greater than 100% compressive strain is impossible. Therefore. the contact penetration
(i.e. cartilage compression) term was scaled by an exponent (exp=9)[105]. This caused
the contact force to rise asymptotically at a penetration of 2.6 mm, which corresponded to
a compressive strain of 86.7% for a cartilage thickness of 3mm (within the range of
cartilage thickness at the ankle and subtalar joints).
The ADAMS dynamic simulation software was used to assemble the dynamic model,
incorporate the geometric and structural properties described above, apply boundary
conditions that mimicked the experimental ones and then generate and solve the dynamic
equations of motion under the various forcing functions. The default integrator (GSTIFF,
SI1) was used to formulate and solve the equations of motion using an integration step
evaluating the results of their models.
192
size of 0.01 and an integrator error of 0.001. The RAPID[96] interference detection
algorithm with a default tolerance of 300 was used to model contact between the
articulating surfaces of the bones.
SIMULATIONS
ach model was simulated under cyclic loads of plantarflexion / dorsiflexion, inversion /
ation / external rotation and anterior drawer with all ligaments
f one normal subject obtained using a six-degree-of-
eedom mechanical linkage, the Ankle Flexibility Tester (AFT), 2) the kinematic data of
ed from a stress MR study of the same subject and 3) the
E
eversion, internal rot
included and with the ATFL excluded and the ATFL and CFL both excluded from the
model. The model was also tested under static loads in the movements of inversion,
anterior drawer, plantarflexion and dorsiflexion. The static inversion and anterior drawer
tests were simulated with and without ATFL and combined ATFL and CFL ligament
components.
Following model development and simulation, the model output was compared to three
types of independent experimental data: 1) the experimental ankle joint complex
flexibility and range of motion data o
fr
the hindfoot joints obtain
experimental kinematic data of one cadaver in the intact condition and with two
simulated injuries (ATFL sectioned and ATFL + CFL sectioned). These data were
obtained from tests performed on the subject that the model was based. Each model’s
flexibility data, kinematics and ligament loading patterns were also compared to data
from the literature.
193
MODEL EVALUATION
Comparison of Experimental Measurements and Model Predictions
The comparison between experimental and numerical results was encouraging because
the model captured fundamental joint load-displacement characteristics (hysteresis in the
loading-unloading paths, high early flexibility, and a non-linear decrease to low late
flexibility)[60]. However, the comparison of the numerical and experimental results
indicated the weaknesses of the model, which primarily were: underestimation of the
stiffness of the subtalar ligaments and improper load sharing by the collateral ankle joint
ligaments.
Overall, the motions measured in the in vivo sMRI experiments were small and the in
vivo model over-estimated joint motions. Furthermore, the in vivo experimental load-
displacement data did not exhibit high early flexibility and a non-linear rapid decrease in
late flexibility (Figures 28-30) predicted by the model and described previously [60, 112].
These data were affected by soft tissue motion. The surrounding tissues (tendons, skin) of
the patient, which were excluded from the hindfoot model, may act to decrease early
flexibility in vivo. The support provided to the joints by the tendons and surrounding soft
tissues may decrease joint range of motion in vivo and explain, in part, why the model
tends to over-predict in vivo joint motions.
The in vitro model captured the experimental patterns of change in the ankle joint
complex when ligament rupture was simulated. However, the ankle joint contributed to
the majority of the rotations for the in vitro inversion experiment in all conditions. On the
194
contrary, the model predicted that the subtalar joint contributed to the majority of motion
ligaments are too flexible and further understanding of the ligament mechanical
pro
of joint
Sensiti
in the in vitro model. The results indicate that the load-strain characteristics of the
subtalar
perties on a patient-specific basis is necessary for accurate patient-specific predictions
function.
vity Analyses
ults of the sensitivity analyses suggest that the model can predict the effect of
ng the major passive support structures of the joint despite the error introduced
identifying the ligament insertions or representing their geometry with linear
ts. Kinematic changes caused by altering model condi
The res
rupturi
when
elemen tions (i.e. removing
liga
represe
in the C
and CF
239.4%
Int
ments) are much larger than the changes introduced into the model by error in
nting ligament geometry. For example, much smaller kinematic changes occurred
FL sensitivity analysis (≤ 10.91%). than what occurred when removing the ATFL
L compared to the ATFL removed condition (477% change in vivo (Table34),
change in vitro (Table 36)).
er-model Variability
atient-specific models incorporated the morphological differences between
ens and, as in the experiments, the hindfoot’s mechanical response to loading was
The p
specim
sensitive to these differences. These results are encouraging because they indicate that the
mo
ligamen
del may be sufficiently sensitive to changes in input parameters (joint geometry,
t orientation, ligament length) to predict changes in model output, on a patient-
195
specifi
subtala
propert
numero
answer
Sim
biologi
the res
ability
parame oving ligaments) in hindfoot mechanics.
c basis, given an appropriate level of model refinement. (i.e. better knowledge of
r ligament structural properties, knowledge of patient-specific ligament
ies). This suggests that this model may eventually be a tool for addressing
us clinical questions on a patient-specific basis, such as surgical optimization; or
ing engineering design problems, including development of joint prostheses.
ilar to an experimental study, it is necessary to test a group of models, which reflects
cal variability within the human population and to perform statistical analyses on
ults. This will allow statistically significant conclusion to be made regarding the
of the model to describe foot mechanics and predict the effects of changing model
ters (i.e. rem
196
AS The m
below:
1)
ped. This is a fundamental problem in
fication of changes between
conditions. The problem can be addressed in vitro by developing a heel-foot plate
rigidly locks the heel to the foot plate.
echanical properties of the interosseous talocalcaneal ligament and the
riate.
3) The material properties of the collateral ligaments can deviate substantially from
published data[105], indicates that cartilage has a higher modulus under the more
SUMPTIONS AND LIMITATIONS
odel was based on several assumptions and limited in several ways as described
Rotations occur in the experiment that the model does not predict because the
experimental ankle joint complex constraints differed from the modeled
constraints. These constraints differed because the skin deformed and the fixation
points locking the heel to the foot plate slip
vivo, but can be partially addressed by performing a consistent foot fixation
protocol. This allows for more consistent quanti
interface system that
2) The m
cervical ligament are undocumented; therefore their load-strain characteristics
were estimated by scaling the ATFL’s properties by the ITCL’s and the CL’s
respective estimated insertion areas. The experimental comparison indicated that
the model over-estimated motion at the subtalar joint; therefore this assumption
may be inapprop
their average values[39], therefore generalized load-displacement properties for
the ligaments[102] may be inadequate for developing patient-specific predictions
of joint function.
4) We may have underestimated the contact stiffness in our model because recently
197
physiologically realistic dynamic unconfined compression loading conditions.
The modulus we based our stiffness on was derived from equilibrium confined
compression tests[104, 106], which are less representative of cartilage’s
iological loading environment[105]. These new data also concluded that
ar joint is constrained by the ligaments,
therefore it is possible that the gaps will change subtalar mechanics, particularly
lcaneal articulation, where there is only a small area of
of the model. Under
this experiment’s near quasi-static loading conditions (≤ 3 s loading times), the
damping term would not drastically effect model mechanics as shown through a
contact damping coefficient sensitivity study in which large variations in this term
caused small increases in oscillations in the hindfoot’s load-displacement curves.
phys
cartilage rarely exceeded 20% compressive strain even under the most strenuous
activities; therefore, the exponent term used in this study was too low.
5) The hindfoot’s distal structures, including the bones (talus and navicular) and soft
tissues (spring ligament), were assumed to have minimal effect on hindfoot
mechanics if no axial loads were applied to the foot. The talo-navicular joint is
highly flexible and therefore will provide minimal resistance to joint motion while
unloaded.
6) It is unlikely that the inter-cortical gaps will cause dramatic changes in ankle joint
range of motion and flexibility[32] because the talus is highly constrained on 3
sides by the tibia and fibula. The subtal
at the anterior talo-ca
contact[67].
7) The contact damping coefficient was chosen to be small (1.0 N*s/mm in vivo, 0.1
N*s/mm in vitro) so that it did not dominate the dynamics
198
8) Sensitivity analysis of the representation of the TSL and TCL structure indicated
nt this ligament with proximal and distal
components, instead of just 1 at its center.
that small changes occurred in hindfoot kinematics when representing them by
greater than a total of 3 elements, therefore, it may be adequate to represent these
structures with 3 elements. Large changes in ankle joint kinematics occurred after
increasing the number of deep PTTL components from 1 to 2 structures; therefore
it may be more appropriate to represe
9) Rigid constraint of the fibula will not influence the results because only small
movements occur in the fibula during foot motion.
MODEL PREDICTIONS
Kinematics
Like previous experiments, the models predicted that the ankle joint and the subtalar joint
contributed to the entire motion across the ankle joint complex[53] in all measured
motions (inversion / eversion, plantarflexion / dorsiflexion, anterior drawer). Unlike
previous experiments[53], the subtalar joint contributed to substantially more of the
rotation in both models
Like the results of previous experiments[53], the in vitro and in vivo models predicted
at the movements of dorsiflexion / plantarflexion occur both at the ankle joint and the
subtalar joint, but primarily at the ankle joint. Both the in vitro and in vivo models greatly
overestimated the amount of ankle joint dorsiflexion compared to an in vivo study of 18
th
199
subjects [112]; therefore, th ligaments constraining the e load-displacement properties of
hindfoot in dorsiflexion may be too flexible.
Flexibility
Like previous experimental studies, the ankle joint complex of the in vitro and in vivo
models had highly non-linear load-displacement properties in all directions[53, 60, 112].
Unlike previous in vitro experimental studies[60], both models predicted a drastic
decrease in late flexibility compared to early flexibility. Two factors may influence the
predicted low late flexibility: 1) the non-linear exponential formulation used to describe
e ligament load-strain characteristics and 2) in the experiments, soft tissue deformation
ge of motion and therefore increase late flexibility. Lack of
th
may increase apparent ran
patient-specific knowledge of the hindfoot ligament mechanical properties may
contribute to the disagreement between the early flexibility measured in the models and
presented in previous experimental studies[60].
Ligament Loading Patterns
The models overpredicted the strain in several ligaments (ATFL, CFL, TCL) compared to
previous experimental data [122] (Table 51). This indicates that the load-strain
rmulations used to describe these structures may not be accurate. fo
200
PRELIMINARY CLINICAL SIGNIFICANCE
The hin
dfoot models capture the fundamental load-displacement characteristics
ysteresis in the loading-unloading paths, high early flexibility, and a non-linear
re sensitive to changes in bone
eometry because both measurements differed between the two models, which contained
two unique sets of bone geometries. This indicates that the model may be useful for
assessing the effects of implant geometry and fusion on ligament load sharing and
hindfoot kinematics. The ligament load-sharing and joint kinematics were sensitive to
removal of the ATFL and CFL; therefore the model may be effective in assessing how
surgical and conservative treatments restore these characterisitics in the hindfoot with
ligament injuries. The load-displacement data predicted by the model may also be used to
identify more sensitive and specific manual methods for ligament injury diagnosis. One
could parametrically determine the direction in which joint displacement increases the
ost after removal of a ligament element from the model. These could be targeted as
(h
decrease to low late flexibility) exhibited by all joints[29, 60]. The results indicate that
hysteresis in the neutral region and low late flexibility depend on both contact of the
articulating surfaces and ligament constraints.
The predicted ligament forces and joint kinematics we
g
m
potential diagnostic tests.
201
CHAPTER 7. FUTURE WORK
address some odeficiencies in the existing model,
suc hich
wo cal
dec
reconst int fusion procedures
as
dynami
SHOR Fut rger group of models (currently
RI data for n = 8 cadavers exists) from in vitro and in vivo test specimens so that the
utput data has sufficient statistical power. The larger database of models would
incorporate inter-subject variability and is therefore essential for making statistically
significant conclusions about the model’s ability to predict hindfoot mechanics. Just as in
experimental studies, it is impossible to draw meaningful conclusions from a small data
set.
The results from the present models indicate that a better understanding of the subtalar
ligaments’ (ITCL, CL) mechanical properties is necessary. In addition, several subtalar
ligaments, which were excluded from the present model, such as the lateral talocalcaneal
ligament, the fibulotalocalcaneal ligament, inferior extensor retinaculum, may need to be
Future work in the development of the hindfoot model can be divided into 2 focus areas:
1) short term projects, which would
h as parameter identification and sample size; 2) long-term areas of focus, w
uld consider future applications of the hindfoot model (i.e. its use for making clini
isions in areas such as ligament injury diagnosis and planning ligament
ructions, and designing ankle replacement and optimizing jo
described above) and further extension of it to incorporate the entire foot or the
c components (muscles and tendons) of the lower limb.
T-TERM GOALS
ure work must focus on developing and evaluating a la
sM
o
202
mechanically tested and if they h tural properties, included in the
odel.
calculating the ligament structural properties on a patient-specific basis, considering the
large standard deviation in ligament mechanical properties[39]. Furthermore, if tests
occurred over larger time periods (>6 s), a relaxation term should be included because the
74 MPa) under the more physiological loading conditions of
also rarely undergoes strain of greater than 20% [105] therefore, the exponent term in the
st also be increased to reflect this in future models. Furthermore,
viously published cartilage loading data.
It is also necessary to perform additional sensitivity analyses on model parameters,
nalyses should explore the
changes in bone volume on model output (i.e. how does exclusion of cartilage geometry
ave substantial struc
m
It may be necessary to develop experimental and numerical methods for measuring and
ligament forces can decrease by greater than 15%[102].
Future studies must explore the effects of increasing the contact stiffness term because
recently published studies[105] indicated that cartilage’s elastic modulus is higher than
that used in this study (0.3
unconfined cyclic compression tests (Maximum Modulus = 65.7 MPa at 1 Hz). Cartilage
contact formulation mu
in order to obtain an improved estimate of contact damping, a simple viscoelastic model
could be matched to pre
particularly the contact parameters mentioned above. The a
effect of variations in contact stiffness, contact penetration exponent and the effects of
affect joint mechanics?).
203
LONG-TERM GOALS Once the existing model has been sufficiently evaluated and confirmed, it is possible to
1) Studying ankle joint fusion, subtalar joint fusion and total ankle replacement
kinematics and mechanics
the articulating surfaces
igament reconstruction hich incorporate the ma
altered ligament orientations
navicular and related soft tissues
extend it in various ways including:
2) Studying of the effects of weight-bearing on model
3) Developing of finite element analysis of the hindfoot structures in order to
determine the stress distribution at
4) Studying of l surgeries, w terial
properties of the replacement materials and the
5) Modeling the distal structures of the foot, beginning at the level of the cuboid,
6) Modeling the tendons for future gait simulations.
204
List Of References
[2] S Serrate-Sztein B Mittleman, JH Klippel, et al., “Questions and Answers about
[3] “Rheumatoid Arthritis of the Foot and Ankle,” 2004,
[4] W Miehlke N Gschwend, P Rippstein, BR Simmen, “Compression Arthrodesis of
[5] JG Kennedy JA Harty, K Casey, W Jan, WB Quinlan, “Outcome After Single
[6] B Hintermann V Valderrabano, “Total ankle replacement,” Foot Ankle Clin N
nal Course Lect, vol. 51, pp. 117-181, 2002.
Complex during Dorsiflexion/Plantarflexion,” Foot Ankle Int, vol. 16, pp. 633-636, 1995.
s and Related Research, pp. 49-55, 1991.
[11] MT Pyevich CL Saltzman, JJ Callaghan, FG Alvine, “Total Ankle Artroplasty: A
ion Part I. Replacement Systems, Indications, Contraindications,” Am J Orthop, pp. 604-609, 2000.
[1] “The Facts about Arthritis,” 2004, http://www.arthritis.org/resources/gettingstarted/default.asp, Arthritis Foundation, 2000.
Arthritis Pain,” , vol. 2004, http://www.niams.nih.gov/hi/topics/arthritis/arthpain.htm#1, 2001.
http://www.footcaremd.com/fc_a_rheumatoidarthritis.html American Academy od Orthopaedic Surgoens, 2001.
the Rheumatoid Ankle and Hindfoot,” Clin Orthop Rel Res, vol. 1, pp. 75-86, July 1997.
Technique Ankle Arthrodesis in Patients with Rheumatoid Arthritis,” Clin Orth Rel Res, vol. 1, pp. 131-138, 2003.
Am, vol. 8, pp. 375-405, 2003.
[7] LH Gill, “Principles of Joint Arthroplasty,” AAOS Instructio
[8] B Hintermann BM Nigg, S Nat, “Influence of Arthrodeses on Kinematics of the Axially Loaded Ankle
[9] Mann RA Van Manen JW, Wapner K, Martin J, “Ankle Fusion,” Clinical Orthopaedic
[10] SK Neufeld TH Lee, “Total Ankle Athroplasty: Indications,Results and Biomechanical Rationale,” Am J Orthop, pp. 593-602, 2000.
Unique Design,” J Bone Joint Surg, vol. 80-A, pp. 1410-1420, 1998.
[12] FG Alvine RA Mann RW Sanders, AK Walling, “Total Ankle Replacement: A Surgical Discuss
205
[13] FG Alvine RA Mann RW Sanders, AK Walling, “Total Ankle Replacement: A Surgical Discussion Part II: The Clinical and Surgical Experience,” Am J Orthop, pp. 675-679, 2000.
[14] CL Saltzman TE McIff, JA Buckwalter, TD Brown,, “Total Ankle Replacement Revisited,” J Orthop Sport Phys, vol. 20, pp. 56-67, 2000.
[15] PM Boruta JO Bishop, WG Braly, “Acute lateral ankle ligament injuries: a literature review,” Foot Ankle, vol. 11, pp. 107-113, 1990.
[16] “The Foot and Ankle,” 2004, http://orthoinfo.aaos.org/fact/thr_report.cfm?thread_id=100&topcategory=foot American Academy of Orthopaedic Surgoens, 2001.
[17] “Sprained Ankle,” 2004, http://orthoinfo.aaos.org/fact/thr_report.cfm?thread_id=152&topcategory=foot), American Orthopaedic Foot and Ankle Society, 2001.
[19] B Hintermann A Boss, D Schafer, “Arthroscopic Findings in Patients with
[20] KD Harrington, “Degenerative arthritis of the ankle secondary to long-standing
[21] JM Meyer J Garcia, P Hoffmeyer, D Fritschy, “The Subtalar Sprain. A
amagata M, Tamaki T, “Influence of the interosseous talocalcaneal ligament injury on stability of the ankle-subtalar joint complex - a cadaveric
elsen, “Seven years follow-up after ankle inversion trauma,” Scand J Med Sci Sports, vol. 12, pp. 129-135, 2002.
2000.
nts: Investigations of the Anterior Drawer and Adduction Maneuvers in Autopsy Specimens,” Clin Orthop Rel Res, vol. 266, pp. 170-179, 1991.
[18] DT Keefe SL Haddad, “Subtalar instability. Etiology, dignosis and management,” Foot Ankle Clin N Am, vol. 7, pp. 577-609, 2002.
Chronic Ankle Instability,” Am J Sports Med, vol. 30, pp. 402-409, 2002.
lateral ligament instability,” J Bone Joint Surg Am, vol. 61, pp. 354-361, 1979.
Roentgenographic Study,” Clin Orthop Rel Res, pp. 169-173, 1988.
[22] Tochigi Y Takahashi K, Y
experimental study,” Foot Ankle Int, vol. 21, pp. 2000, 2000.
[23] L Konradsen L Bech, M Ehrenbjerg, T Nick
[24] T Fujii Z-P Luo, HB Kitaoka, K.N An, “The manual stress test may not be sufficient to differentiate ankle ligament injuries,” Clin Biomech, vol. 15, pp. 619-623,
[25] Kjaersgaard-Andersen P Frich LH, Madsen F, Helmig P, Sogard P, Sojbjerg JO, “Instability of the Hindfoot After Lesion of the Lateral Ankle Ligame
206
[26] Rasmussen O, “Stability of the Ankle Joint. Analysis of the function and Traumatology of the Ankle Ligaments,” Acta Orthop Scand, vol. 56, pp. 1-61, 1985.
[27] Johannsen A, “Radiological Diagnosis of Lateral Ligament Lesion of the Ankle,”
[28] R Bahr F Pena, J Shine, WD Lew, C Lindquist, S Tyrdal, L Engebretsen,
op Scand, vol. 68, pp. 435-441, 1997.
9-614, 1995.
-244, 2001.
Acta orthop Scand, vol. 52, pp. 99-102, 1981.
ports Med, vol. 13, pp. 295-300, 1985.
oach, “Mechanics of the ankle and subtalar joints revealed through a 3D quasi-static stress MRI technique,” J Biomech, vol. In Press, 2004.
nt of an osseo-ligamentous finite element model of the human ankle joint,” Int J Solids Struct, vol. 38,
ari, “A Preliminary Numerical Model of the Human Ankle under Impact Loading,” presented at PLEI Conference, Washington
ormable Bone,” presented at 41st Stapp Car Crash Conference, Orlando (FL), 1997.
e-dimensional anatomically detailed foot model: A foundation for a finite element
at
Acta Orthop Scand, vol. 49, pp. 295-301, 1978.
“Mechanics of the anterior drawer and talar tilt tests: A cadaveric study of lateral ligament injuries of the ankle,” Acta Orth
[29] H Tohyama BD Beynnon, PA Renstrom, MJ Theis, BC Fleming, MH Pope, “Biomechanical Analysis of the Ankle Anterior Drawer Test for Anterior Talofibibular Ligament Injuries,” J Orthop Res, vol. 13, pp. 60
[30] W Liu S Siegler, L Techner, “Quantitative measurement of ankle passive flexibility using an arthrometer on sprained ankles,” Clin Biomech, vol. 16, pp. 237
[31] Rasmussen O Tovborg-Jensen I, “Anterolateral Rotational Instability in the Ankle Joint,”
[32] DM Stormont BF Morrey, K-N An, JR Cass, “Stability of the loaded ankle. Relation between articular restraint and primary and secondary restraint,” Am J S
[33] S Siegler JK Udupa, SI Ringleb, CW Imhauser, BE Hirsch, D Odhner, PK Saha, E Okereke, N R
[34] FA Bandak RE Tannous, T Toridis, “On the developme
pp. 1681-1697, 2001.
[35] M Beaugonin E Haug, G Munck, D Ces
D.C., 1995.
[36] M Beaugonin E Haug, D Cesari, “Improvement of Numerical Ankle/Foot Model: Modeling of Def
[37] DLA Camacho WR Ledoux, Rohr ES, BJ Sangeorzan BJ, RP Ching, “A thre
simul ion and means of quantifying foot-bone position,” J Rehabil Res Dev, vol. 39, pp. 401-410, 2002.
207
[38] W Ledoux R Ching , ES Rohr, B Sangeorzan, “The Development and Validation of a Computational Foot and Ankle Model,” presented at 22nd Annual EMBS International Conference, Chicago (IL), 2000.
oint,” Foot Ankle, vol. 8, pp. 234-242, April 1988.
finite element analysis,” Clin Biomech, vol. 16, pp. 614-620, 2001.
3D two arch model of a normal human foot,” Mech Res Commun, vol. 23, pp. 387-393, 1996.
odeling and Analysis of Stresses in Normal and Early Stage Hansen's Disease with Muscle Paralysis,” J Rehabil Res Dev, vol. 36, pp. 252-263, 1999.
[43] LA Setton DM Elliott, VC Mow, “Altered mechanics of cartilage with
[44] JK Horstman GS Kantor, KM Samuelson, “Investigation of Lateral Ankle e
parison of MRI and Clinical Examination of Acute Lateral Ankle Sprains,” Foot Ankle, vol. 17, pp. 533-537, 1996.
in the diagnosos of ruptures of the lateral ligaments,” Am J Sports Med, vol. 19, pp. 583-587, 1991.
ors for post-traumatic lateral instability of the ankle?,” Ergonomics, vol. 43, pp. 1718-1725, 2000.
sch, E Eils, S Neller, S Benesch, A Herbst, D Rosenbaum, L Claes, “Reconstruction of the lateral ligaments: do the anatomical
[49] D Rosenbaum HP Becher, HJ Wilke, LE Claes, “Tenodeses destroy the kinematic
-B, pp. 162-168, January 1998.
[39] S Siegler J Block, CD Schneck, “The Mechanical Characteristics of the Collateral Ligaments of the Human Ankle J
[40] W-P Chen F-T Tang, C-W Ju, “Stress distribution of the foot during mid-stance to push-off in barefoot gait: a 3-D
[41] S Jacob LH Braak, A Huson, “Stresses in a
[42] S Jacob, “Three-dimensional Foot M
osteoarthritis: human osteoarthrits and an experimental model of joint degeneration,” Osteoarthr Cartilage, vol. 7, pp. 2-14, 1999.
Ligam nt Reconstruction,” Foot Ankle, vol. 1, pp. 338-342, 1981.
[45] C Frey J Bell, L Teresi, R Kerr, K Feder, “A com
[46] EFC Verhaven M Shahabpour, “The accuracy of three-dimensional magnetic resonance imaging
[47] E Barbaix P Van Roy, JP Clarys, “Variations of anatomical elements contributing to subtalar joint stability: intrinsic risk fact
[48] R Schmidt E Cordier, C Bert
procedures restore physiologic ankle kinematics,” Foot Ankle Int, vol. 25, pp. 31-6, 2004.
coupling of the ankle joint complex: A Three-Dimensional In Vitro Analysis of Joint Movement,” J Bone Joint Surg, vol. 80
208
[50] N Usami S Inokuchi, E Hiraishi, M Miyanaga, A Waseda, “Clinical Application of Artificial Ligament for Ankle Instability - Long-Term Follow-Up,” J Long-Term F Med, vol. 10, pp. 239-250, 2000.
[52] J Karlsson BI Eriksson, PA Renstrom, “Subtalar Ankle Instability. A Review,”
[53] Chen J, “An Investigation on the Three Dimensional Mechanical Characteristics
[55] A Lundberg OK Svensson, C Bylund, I Goldie, G Selvik, “Kinematics of the
[56] A Lundberg OK Svensson, C Bylund, G Selvik, “Kinematics of the Ankle /Foot le
[57] B Hintermann BM Nigg, “In Vitro Kinematics of the Axially Loaded Ankle le
[58] GJ Sammarco AH Burstein, VH Frankel, “Biomechanics of the Ankle: A a
-N An, “Physiological Elongation of Ligamentous Complex Surrounding the Hindfoot Joints: In Vitro Biomechanical Study,”
Lapointe S Siegler, Hillstrom H, RR Nobilini RR, A Mlodzienski, L Techner, “Changes in the Flexibility Characteristics of the Ankle Complex Due to Damage to the
[61] BM Nigg G Skarvan, CB Frank, MR Yeadon, “Elongation and Forces of Ankle e
mig, K Sobald, “The stabilizing effect of the ligamentous structures in the sinus and canalis tarsi on mevements in the hindfoot. An experimental study,” Am J Sports Med, vol. 16, pp. 512-516, 1988.
[51] Reed TM, “Modified Elmslie Lateral Ankle Stabilizatin Procedure,” J Foot Ankle Surg, vol. 35, pp. 162-191, 1996.
Sports Med, vol. 24, pp. 337-346, 1997.
of the Ankle and Subtalar Joints,” in Mechanical Engineering and Mechanics. Philadelphia: Drexel University, 1989, pp. 131.
[54] A Lundberg I Goldie, B Kalin, G Selvik, “Kinematics of the Ankle/Foot Complex: Plantarflexion and Dorsiflexion,” Foot Ankle, vol. 9, pp. 194-200, 1989.
Ankle/Foot Complex-Part2: Pronation and Supination,” Foot Ankle, vol. 9, pp. 248-253, 1989.
Comp x-Part 3: Influence of Leg Rotation,” Foot Ankle, vol. 9, pp. 304-309, 1989.
Comp x in Response to Dorsiflexion and Plantarflexion,” Foot Ankle Int, vol. 16, pp. 514-518, 1995.
Kinem tic Study,” Orthop Clin N Am, vol. 4, pp. 75-95, January 1973.
[59] Z-P Luo HB Kitaoka, H-C Hsu, H Kura, K
Foot and Ankle Int, vol. 18, pp. 277-283, 1997.
[60] SJ
Lateral Collateral Ligaments: An in Vitro and In Vivo Study,” J Orthop Res, vol. 15, pp. 331-341, 1997.
Ligam nts in a Physiological Range of Motion,” F and A Int, vol. 11, pp. 30-40, 1990.
[62] P Kjaersgaard-Andersen J-O Wethelund, P Hel
209
[63] GA Knudson HB Kitaoka, C-L Lu, ZP Luo K-N An, “Subtalar joint stability Talocalcaneal interosseous ligament function studied in cadaver specimens,” Acta Orthop Scand, vol. 68, pp. 442-446, 1997.
[66] AP Boss B Hintermann, “Anatomical Study of the Medial Ankle Ligament le
[67] SK Sarrafian, Anatomy of the foot and ankle : descriptive, topographic,
[68] SK Sarrafian, “Biomechanics of the Subtalar Joint Complex,” Clin Orthop Rel
[69] CD Schneck S Siegler, J Block, R Seliktar, M Mesgarzadeh, “The Geometry of
Foot Ankle, vol. 10, pp. 30-35, 1989.
erence, San Diego (CA), 1999.
2002.
[64] O Rasmussen C Kromann-Andersen, S Boe, “Deltoid Ligament,” Acta Orthop Scand, vol. 54, pp. 36-44, 1983.
[65] Harper MC, “Deltoid Ligament: An Anatomical Evaluation of Function,” Foot Ankle, vol. 8, pp. 19-22, 1987.
Comp x,” Foot Ankle Int, vol. 23, pp. 547-553, 2002.
functional. Philadelphia: Lippincott, 1983.
Res, pp. 17-26, 1993.
the Components of the Deltoid Ligament of the Ankle and its Potential Clinical Significance,” unpublished, 1988.
[70] P Kjaersgaard-Andersen J-O Wethelund, P Helmig, K Soballe, “Stabilization of the Tibiocalcaneal Fascicle of the Deltoid Ligament on Hindfoot Movements: An Experimental Study,”
[71] JK Udupa BE Hirsch, HJ Hillstrom, GR Bauer, JB Kneeland, “Analysis of in vivo 3-D Internal Kinematics of the Joints of the Foot,” IEEE Trans Biomed Eng, vol. 45, pp. 1387-1396, 1998.
[72] P Beillas F Lavaste, D Nicolopoulos, K Kayventash, KH Yang, S Robin, “Foot and Ankle Finite Element Modeling Using CT-Scan Data,” presented at 43rd Stapp Car Crash Conf
[73] EP Salathe Jr. GA Arangio EP Salathe, “A Biomechanical Model of the Foot,” J Biomech, vol. 19, pp. 989-1001, 1986.
[74] Salathe EP, “A Biomechanical Model of the Foot: The Role of Muscles, Tendons, and Ligaments,” J Biomech Eng, vol. 124, pp. 281-287,
[75] A Leardini F Catani, S Giannini, “A geometric model of the human ankle joint,” J Biomech, vol. 32, pp. 585-591, 1999.
210
[76] Y Kameyama E Genda, K Yokoe, “Theoretical Stress Analysis of Ankle Sprain,” presented at 50th Annual Meeting of the Orthopaedic Research Society, San Francisco, CA, 2004.
[77] PG Bedewi KH Digges, “Investigating Ankle Injury Mechanisms in Offset Frontal Collisions Utilizing Computer Modeling and Case-Study Data,” presented at 43rd Stapp Car Crash Conference, San Diego (CA), 1999.
e phase of walking.,” J Biomech, vol. 26, pp. 1091-1104, 1993.
[80] P Klein S Mattys, M Rooze, “Moment arm length variations of selected muscles
[81] F Corazza JJ O'Connor, A Leardini, V Parenti Castelli, “Ligament fibre
[82] BE Hirsch JK Udupa, E Stindel, “Tarsal Joint Kinematics Via 3D Imaging,” Crit
ine phase contrast magnetic resonance imaging to non-invasively study in vivo knee dynamics,” J Biomech, vol. 31, pp. 21-26,
arma, BE Hirsch, RA Lotufo, “User-Steered Image Segmentation Paradigms: Live Wire and Live Lane,” Graphical Models
[85] PK Saha JK Udupa, AX Falccao, BE Hirsch, S Siegler, “Iso-shaping rigid bodies
[86] JK Udupa D Odhner, S Samarasekara, RJ Goncalves, K Iyer, K Venugopal, S
[87] E Stindel JK Udupa, BE Hirsch, D Odhner, “A characterization of the geometric
[78] SH Scott DA Winter, “Biomechanical model of the human foot: kinematics and kinetics during the stanc
[79] S Scott DA Winter, “Biomechanical Model of the Human Foot: Kinematics and Kinetics During the Stance Phase of Walking,” J Biomech, vol. 26, pp. 1091-1104, 1993.
acting on the talocrural and subtalar joints during movement: an in vitro study,” J Biomech, vol. 29, pp. 21-30, 1996.
recruitment and forces for the anterior drawer test at the human ankle joint,” J Biomech, vol. 36, pp. 363-372, 2003.
Rev Diag Imag, vol. 41, pp. 403-449, 2000.
[83] FT Sheehan FE Zajac, JE Drace, “Using c
1998.
[84] AX Falcoa JK Udupa, S Samarasekera, S Sh
and Image Processing, vol. 60, pp. 233-260, 1998.
for estimating their motion from image sequences,” IEEE Trans Med Im, vol. 23, pp. 63-72, 2004.
Furuie, “3DVIEWNIX: An open transportable, multidimensional, multiparametric imaging software system,” presented at Proceedings of SPIE, 1994.
architecture of the peritalar joint complex via MRI: an aid to the classification of foot type,” IEEE Trans Med Im, vol. 18, pp. 753-763, 1999.
211
[88] E Stindel JK Udupa, BE Hirsch, D Odhner, C Couture, “MR image analysis of the morphology of the rearfoot: application to classification of bones,” Comput Med Imag Graph, vol. 23, pp. 75-83, 1999.
CA, 1987.
%20marching%20cubes%20algorithm, 1998.
f Mechanical Engineering, University of Michigan, Ann Arbor, ME 543 March 22, 2001 2001.
46-760.
[95] UM Ascher LR Petzold, Computer Methods for Ordinary Differential Equations
ce on Computer Graphics and Interactive Techniques, 1996.
sity of Wisconsin, Fall 2001.
[98] S Chenney, “CS 638: Computer Gaming technology,” , vol. 2004, http://www.cs.wisc.edu/~schenney/courses/cs638-f2001/ ed, 2001.
[99] SP Raya JK Udupa, “Shape-based Interpolation of Multidimensional Objects,” IEEE Trans Med Imag, vol. 9, pp. 32-42, March 1990.
[100] R Fisher S Perkins, A Walker and E Wolfart., “Gaussian Smoothing,” , vol. 2004, http://homepages.inf.ed.ac.uk/rbf/HIPR2/gsmooth.htm ed.
[89] WE Lorensen HE Cline, “Marching Cubes: A High Resolution 3D Surface Construction Algorithm,” presented at SIGGRAPH, Anaheim,
[90] O'Rourke J, “Computer Graphics Algorithms Frequently Asked Questions,” 2004, http://www.exaflop.org/docs/cgafaq/What%20is%20the
[91] D Negrut B Harris, “ADAMS Theory in a Nutshell,” Department o
[92] “ADAMS/Solver Manual,” in Debug: MSC Software Corporation, 2003, pp. 7
[93] “ADAMS/Solver Manual,” in Integrator: MSC Software Corporation, 2003, pp. 218-259.
[94] CW Gear, “Simultaneous Numerical Solution of Differential-Algebraic Equations,” IEEE Trans Circ Theory, vol. 18, pp. 89-95, 1971.
and Differential-Algebraic Equations. Philadelphia: Society for Industrial and Applied Mathematics, 1998.
[96] S Gottschalk MC Lin, D Manocha, “OBB Tree: A Hierarchial Structure for Rapid Interference Detection,” presented at The 23rd Annual International Conferen
[97] S Chenney, “CS 638: Computer Game Technology,” , vol. 2004, http://www.cs.wisc.edu/~schenney/courses/cs638-f2001/ ed: Univer
212
[101] CD Schneck M Mesgarzadeh our, GJ Ross, “MR Imaging of the ost Commonly Injured Ankle Ligaments Part 1. Normal Anatomy,” Radiol, vol. 184,
p. 499-506, 1992.
02] JR Funk GW Hall, JR Crandall, WD Pilkey, “Linear and Quasi-Linear Biomech Eng, vol. 122, pp. 15-22,
Feb 2000.
[103] YC Fung, Biomechanics. Mechanical Properties of Living Tissues. New York:
, “Comparison of Human Humeral Head and Glenoid ,”
r
, 2004.
GA Ateshian, VC Mow, “Comparison of Isotropic and operties from Confined Compression and Indentation,”
ng Conference, 2001.
new Method to Determine the Tensile the Indentation Test,” presented at Orthopaedic rch 2000.
ini, “ISB recommendation on definitions of joint rting of human joint motion - part 1: ankle, hip,
.
Body Motion Calculated from Spatiel Co-980.
dinate System for the Clinical Description of o the Knee,” J Biomech Eng, vol. 105, pp.
, Berman AT, “A six degree of freedom mechanical linkage for measuring the flexibility characteristics of the ankle joint
1996.
[112] S Siegler D Wang, E Plasha, AT Berman, “Technique for In Vivo Measurement f the Three-Dimensional Kineamtics and Laxity Characteristics of the Ankle Joint
Complex,” J Orthop Res, vol. 12, pp. 421-431, 1994.
, A BonakdarpMp
[1Viscoelaatic Characterization of Ankle Ligaments,” J
Springer-Verlag, 1993.
[104] TR Gardner VC MowArticular Cartilage Material Properties Using Transversely Isotropic Biphasic FEM
g Conference, 2001. presented at ASME Bioengineerin
[105] S Park CT Hung, GA Ateshian, “Mechanical response of bovine articulacartilage under dynamic unconfined compression loading at physiological stress levels,”Osteoarth Cart, vol. 12, pp. 65-73
[106] TR Gardner EJ Balaguer, PrTransversely Isotropic Material
presented at ASME Bioengineeri
[107] VC Mow PM Good, TR Gardner, “AProperties of Articular Cartilage Using
esearch Society, Orlando, Florida, MaR
[108] S Siegler C Kirtley, A Leardcoordinate of various joints for the repoand spine,” J Biomech, vol. 35, pp. 543-548, 2002
[109] CW Spoor FE Veldpaus, “Rigid ordinates of Markers,” J Biomech, vol. 13, 1
[110] ES Grood WJ Suntay, “A Joint CoorThree-Dimensional Motions: Application t136-144, 1983.
[111] S Siegler Lapointe S, Nobilini R
complex,” J Biomech, vol. 29, pp. 943-947,
o
213
[113] J Karrholm LI Hansson, G Se of the lateral malleolus. A roentgen stereophotogrammetric analysis,” Acta Orthop Scand, vol. 56, pp. 479-483, 1985.
[114] C Imhauser S Siegler, NA Abidi, DZ Frankel, “The effect of posterior tibialis tendon dysfunction on the plantar pressure characteristics and the kineamtics of the arch
69, 2004.
analysis of flatfoot . 447-451, 1998.
n, R Lusskin, “Evaluation of the UCBL Shoe Insert in the s Planus,” Clin Orthop, vol. 82, pp. 45-58, 1972.
Kuei, WH Akeson, “Quasi-Linear Viscoelastic e,” J Biomech Eng, vol. 102, pp. 85-90, 1980.
vaste, “Dynamic stiffness and damping of scillatory displacement under a free mass system,”
SF Viegas, “A Comprehensive Study of .
25-133, 1994.
t plex around the Ankle and Subtalar Joints,” Foot Anle, vol. 13, pp. 130-136, 1992.
-521, 2001.
taneous Strain nce for the Medial and Lateral
Foot Ankle Int, vol. 23, pp. 825-832, 2002.
lvik, “Mobility
and the hindfoot,” Clin Biomech, vol. 19, pp. 161-1
[115] HB Kitaoka Z-P Luo, K-N An, “Three-Dimensional deformity,” Foot Ankle Int, vol. 19, pp
[116] C Mereday CME DolaTreatment of "Flexible" Pe
[117] S L-Y Woo BR Simon, SC roperties of Normal Articular CartilagP
[118] O Izambert D Mitton, M Thourot, F Lauman intervertebral disc using axial oh
Eur Spine J, vol. 12, pp. 562-566, 2003.
[119] JH Calhoun M Eng, F Li, BR Ledbetter,Pressure Distributionin the Ankle Joint with Inversion and Eversion,” Foot Ankle, vol5, pp. 11
[120] MM Stephens GJ Sammarco, “The Stabilizing Role of the Lateral LigamenCom
[121] AM Moseley J Crosbie, R Adams, “Normative data for passive anklelantarflexion-dorsiflexion flexibility,” Clin Biomech, vol. 16, pp. 514p
[122] S Ozeki K Yasuda, K Kaneda, K Yamakoshi, T Yamanoi, “SimulMeasur
igaments of the Ankle,” ement with Determination of a Zero Strain Refere
L
214
Appendix A
t'; mread(inputfile,'\t');
8,:)',inputdat(7,:)',inputdat(5,:)'];
,:)',inputdat(16,:)',inputdat(13,:)'
inv(XTi_MR)*XTa_MR
ia
nTIB=inv(XTi_MR)*P1TibInMRI' P2TIBinTIB=inv(XTi_MR)*P2TibInMRI' P3TIBinTIB=inv(XTi_MR)*P3TibInMRI' PTTL1TibInMRI=[82.4 63.9 -10.8 1]; PTTL2TibInMRI=[73.8 65.4 -11.8 1]; PTTL3TibInMRI=[82.6 65.2 -8.3 1]; PTTL4TibInMRI=[78.9 65.4 -12 1]; PTTL5TibInMRI=[80.1 64.5 -7.8 1]; PTTL6TibInMRI=[79 63.2 -17 1]; PTTL7TibInMRI=[75.8 65.2 -6.2 1]; PTTL8TibInMRI=[71.5 66.1 -15.7 1];
MATLAB M-File for Coordinate Transformations
inputffile=dl
ile='SieglerL03BonePositionsMATLAB.tx
[r c]=size(file); inputdat=file(1:r,1:c); XCa_MR=[inputdat(2,:)',inputdat(3,:)',inputdat(4,:)',inputdat(1,:)']; XCa_MR=[XCa_MR; [0 0 0 1]] a_MR=[inputdat(6,:)',inputdat(XT
XTa_MR=[XTa_MR; [0 0 0 1]] XTi_MR=[inputdat(12,:)',inputdat(10,:)',inputdat(11,:)',inputdat(9,:)']; i_MR=[XTi_MR; [0 0 0 1]] XT
XFi_MR=[inputdat(15,:)',inputdat(14]; XFi_MR=[XFi_MR; [0 0 0 1]] a_Ti=XT
XCa_Ti=inv(XTi_MR)*XCa_MR XFi_Ti=inv(XTi_MR)*XFi_MR ib%T
TibCentInMRI=[inputdat(9,:) 1] TibInMRI=[71.2 16.9 2.6 1]; P1
P2TibInMRI=[72.3 16.8 2.6 1]; P3TibInMRI=[65.1 68.6 -14.5 1];
ntInTIB=inv(XTi_MR)*TibCentInMRI' TibCeP1TIBi
215
Appendix B
MATLAB M-File for Helical A
inputdat=file(1:r,1:c);
%Ankle Joint Complex a 1,:)]']
a ,:)]']
Ca(XCpauPhiAJC=rad2deg(acos(.5*(XCa_Tip1p2(1,1)+XCa_Tip1p2(2,2)+XCa_Tip1p2(3,3)
.5
n ,3)-
nAJC=.5*[XCa_Tip1p2(3,2)-XCa_Tip1p2(2,3);XCa_Tip1p2(1,3)-XCa_Tip1p2(3,1);XCa_Tip1p2(2,1)-XCa_Tip1p2(1,2)]./SinPhiAJC
(13,:
Ta
aPhiAJ=rad2deg(acos(.5*(XTa_Tip1p2(1,1)+XTa_Tip1p2(2,2)+XTa_Tip1p2(3,3)-1))) AJTranMagTransTalwrtTib=(AJTransVec(1)^2+AJTransVec(2)^2+AJTransVec(3)^2)^.5
xis Calculations
inputfile='ADSScATFLcCFLStartclosedgapALLDATA.txt'; file=dlmread(inputfile,'\t'); [r c]=size(file);
XC _Tip1=[[inputdat(2,:);inputdat(3,:);inputdat(4,:)],[inputdat(; XCa_Tip1=[XCa_Tip1; [0 0 0 1]] XC _Tip2=[[inputdat(6,:);inputdat(7,:);inputdat(8,:)],[inputdat(5; XCa_Tip2=[XCa_Tip2; [0 0 0 1]] X _Tip1p2=XCa_Tip2*inv(XCa_Tip1)
a_Tip1p2(1,1)+XCa_Tip1p2(2,2)+XCa_Tip1p2(3,3)-1) se
-1))) AJCTransVec=XCa_Tip2(:,4)-XCa_Tip1(:,4) MagTransCalwrtTib=(AJCTransVec(1)^2+AJCTransVec(2)^2+AJCTransVec(3)^2)^
Si PhiAJC=.5*sqrt((XCa_Tip1p2(3,2)-XCa_Tip1p2(2,3))^2+(XCa_Tip1p2(1XCa_Tip1p2(3,1))^2+(XCa_Tip1p2(2,1)-XCa_Tip1p2(1,2))^2)
magnAJC=sqrt((nAJC(1)^2+nAJC(2)^2+nAJC(3)^2)) pause %Ankle Joint XTa_Tip1=[[inputdat(10,:);inputdat(11,:);inputdat(12,:)],[inputdat(9,:)]']; XTa_Tip1=[XTa_Tip1; [0 0 0 1]]
XTa_Tip2=[[inputdat(14,:);inputdat(15,:);inputdat(16,:)],[inputdat)]']; X _Tip2=[XTa_Tip2; [0 0 0 1]] XT _Tip1p2=XTa_Tip2*inv(XTa_Tip1)
sVec=XTa_Tip2(:,4)-XTa_Tip1(:,4)
216
APPENDIX B
SinPhiAJ=.5*sqrt((XTa_Tip1p2(3,2)-XTa_Tip1p2(2,3))^2+(XTa_Tip1p2(1,3)-XTa_Tip1p2(3,1))^2+(XTa_Tip1p2(2,1)-XTa_Tip1p2(1,2))^2) nAJ=.5*[XTa_Tip1p2(3,2)-XTa_Tip1p2(2,3);XTa_Tip1p2(1,3)-XTa_Tip1p2(3,1);XTa_Tip1p2(2,1)-XTa_Tip1p2(1,2)]./SinPhiAJ magnAJ=sqrt((nAJ(1)^2+nAJ(2)^2+nAJ(3)^2)) pause
%Subtalar Joint Complex XCa_Tap1=[[inputdat(18,:);inputdat(19,:);inputdat(20,:)],[inputdat(17,:)]']; XCa_Tap1=[XCa_Tap1; [0 0 0 1]] XCa_Tap2=[[inputdat(22,:);inputdat(23,:);inputdat(24,:)],[inputdat(21,:)]']; XCa_Tap2=[XCa_Tap2; [0 0 0 1]] XCa_Tap1p2=XCa_Tap2*inv(XCa_Tap1) PhiSTJ=rad2deg(acos(.5*(XCa_Tap1p2(1,1)+XCa_Tap1p2(2,2)+XCa_Tap1p2(3,3)-1))) STJTransVec=XCa_Tap2(:,4)-XCa_Tap1(:,4) MagTransCalwrtTal=(STJTransVec(1)^2+STJTransVec(2)^2+STJTransVec(3)^2)^.5 SinPhiSTJ=.5*sqrt((XCa_Tap1p2(3,2)-XCa_Tap1p2(2,3))^2+(XCa_Tap1p2(1,3)-XCa_Tap1p2(3,1))^2+(XCa_Tap1p2(2,1)-XCa_Tap1p2(1,2))^2) nSTJ=.5*[XCa_Tap1p2(3,2)-XCa_Tap1p2(2,3);XCa_Tap1p2(1,3)-XCa_Tap1p2(3,1);XCa_Tap1p2(2,1)-XCa_Tap1p2(1,2)]./SinPhiSTJ magnSTJ=sqrt((nSTJ(1)^2+nSTJ(2)^2+nSTJ(3)^2)) pause
217
Vita
Carl William Imhauser EDUCATION
Ph.D. Mechanical Engineering, 2004, Drexel University, Philadelphia, PA, Advisor: Sorin Siegler M.S. Mechanical Engineering, 2000, Drexel University, Philadelphia, PA, Advisor: Sorin Siegler B.S. Electrical Engineering, 1997, Temple University, Philadelphia, PA, magna cum laude
HONORS AND AWARDS • Drexel University Graduate Research Award, Department of Mechanical Engineering and Mechanics,
2004 • Teaching Fellowship, Department of Mechanical Engineering and Mechanics, 2003-2004 • German Academic Exchange Research Fellowship (6 months), begins September 2004 • Koerner Fellowship, College of Engineering Award for academic merit, 2002-2003 • International Society of Biomechanics Doctoral Dissertation Grant, 2001 • Finalist, Orthopedic Foot and Ankle Society Goldener Award for Outstanding Paper, 2001 • Nissen Award Nominee for Outstanding Senior NCAA Gymnastics Student-Athlete, 1997 • NCAA Division I National Horizontal Bar Champion, 1996; NCAA All-American, 1995, 1996 • NCAA Academic All-American, 1994-1997 • George Wallace Hayes Memorial Award, Presented to a Temple University Gymnast with outstanding
gymnastic skill and who exemplifies “Wally.” High ideals, courage, enthusiasm, concern for the welfare of his teammates and a humility that uniquely enough is a quiet source of inspiration. 1997
RESEARCH EXPERIENCE Research Assistant, Drexel University, January 1998-August 2004 • Developed a patient-specific dynamic model of the hindfoot • Participated in study, Biomechanics of Foot/Ankle Injuries using 3D Imaging. (US DHHS grant
AR46902) • Redesigned 6 degree-of-freedom mechanical linkage for testing the hindfoot. • Developed an experimental axial and tendon loading system quasi-statically simulate gait. • Designed experiment to investigate ankle joint distraction for ankle arthroplasty implantation. • Advised Temple University medical students investigating glenohumeral joint biomechanics. • Advised/mentored 1 senior design group and 2 undergraduate honors students.
REFEREED JOURNAL PUBLICATIONS Ringleb SI, Siegler S, Udupa JK, Imhauser CW, et al., Bone Morphology and architecture of the ankle and subtalar joints revealed through a quasi-static three-dimensional stress-MRI technique, J Biomech, Accepted for publication, March, 2004. Imhauser CW, Siegler S et al., The Effect of Posterior Tibialis Tendon Dysfunction on the Plantar Pressure Distribution and the Kinematics of the Arch and the Hindfoot. Clin Biomech, 19: 161-169, 2004. Imhauser CW, Abidi NA, et al., Biomechanical Evaluation of the Efficacy of External Stabilizers in the Conservative Treatment of Adult Acquired Flatfoot Deformity. Foot Ankle Int, 22(8): 727-737,2002.
TEACHING EXPERIENCE Adjunct Instructor, Drexel University, Goodwin College of Professional Studies, Winter 2004
LEADERSHIP EXPERIENCE • President, Mechanical Engineering Graduate Student Association, 2001-2002. • Co-founder and secretary, Drexel University Graduate Student Research Council, 2001-2002. • Temple University Men’s Gymnastics Varsity Team Captain, 1995-1997.
INTERNATIONAL INDUSTRY EXPERIENCE • Evaluated metal structures using microscopes, Diehl Metals, Nuremberg, Germany, July 1996.
• Worked in aluminum foundry machining parts, Ohm and Haner, Olpe, Germany, June 1995. VOLUNTEER WORK
• Reading Tutor, Alexander Adaire Elementary School, Philadelphia, PA, Academic Year 2002-2003.